https://doi.org/10.31449/inf.v48i5.5366                                             Informatica 48 (2024) 97-110   97 
 
Edge Detection and Simulation Analysis of Multimedia Images Based 
on Intelligent Monitoring Robot 
Xiaoyan Wang
*
, Ya Li 
School of Computer Science and Technology, Nanyang Normal University,  
Nanyang, Henan, 473061, China 
E-mail: 2016122095@jou.edu.cn 
*
Corresponding author 
Keywords: multimedia vision, virtual reality, reconstruction of image texture enhancement related search  
Received: October 25, 2023 
Since the problems of imperfect performance and long time consuming occurs to the multimedia image 
edge detection according to the current image edge detection method, which makes the multimedia 
image quality after edge detection not high, in order to remove the edge features contained in 
multimedia images efficiently and solve the defects of ignoring the texture of multimedia images in the 
traditional multimedia image edge detection method, an intelligent monitoring robot-based multimedia 
image edge detection method is put forward in this paper. The method mainly focuses on the similarity 
characteristics of pixels and edge detection points between adjacent regions of multimedia images, and 
uses an iterative method to weight adjacent regions of multimedia images, so that the nature of 
multimedia images is different from the previous traditional images, and the improvement of the viewing 
mode is loved by the audience, who can experience the immersive state and become the subject of the 
image when watching the movie.The VR images are corrected by binocular offset positioning and the 
noise is removed from the corrected images, and the 3D edge detection correlation retrieval method is 
used to obtain the ensemble and outlier values, obtain the maximum/minimum, find outand correct the 
errors, and calculate the value for completion.The final results of the experiment suggest that 
multimedia image edge features can be removed by intelligent monitoring robot effectively withthe 
optimal multimedia image edge effect at ahigh multimedia image processing speed. 
Received: Razvita je metoda za zaznavanje robov multimedijskih slik z uporabo robota za inteligentno 
nadzorovanje, ki izboljša kakovost slik z učinkovitim odstranjevanjem robov. 
 
1   Introduction 
As a virtual environment created by digital technology 
[1-2], virtual reality (VR) is a virtual situation and an 
imaginary world not existing in the real world, and also 
brings customers multiple sensory experiences such as 
sight, sound, touch and smell.It offers brand-new 3D 
perceptual experiences [3-4] to customers, who can 
experience the virtual world through display or data glove 
tools [5].Multimedia image edge detection computing 
creates a novel visual world for customers, and this 
approach is currently applied to several fields and is a new 
development trend.Conventional forms of viewing are 
experienced by the audience following the camera, and 
the director controls the creation of works.Totally 
different from the conventional viewing form (passive 
reception of information), multimedia imagesare a 
game-changer (active reception), and the audience no 
longer needs to experience by following the camera. 
Instead, they can justgo by favorite free conversion, so 
everyone's visual perception and experience of the impact 
of VR is different. Hence, in the film and television 
industry, multimedia images are an irresistible trend in 
the future, which will impress the audience with new 
experiences based on new technical features and viewing 
models and facilitate the interaction of images with 
viewers [6-7]. If language is the traditional tool of 
communication, in multimedia images, the sense of 
hearing and touch cannot be controlled by the media. 
Multimedia images represent a connection between the 
real and virtual worlds by bringing the sensibility close to 
nature and concrete physical objects, rather than using the 
symbolism of symbols to understand the expression of the 
work. Another way of expression is that VR does not use 
words nor needs explanation.The customer uses the 
algorithm of edge detection of multimedia images to reach 
the virtual world and experience the visual, auditory, and 
tactile sensations [8-9]. In the multimedia image, the 
audience is the first to experience the visual and auditory 
senses; multimedia images are noticed because of the 

98   Informatica 48 (2024) 97–110                                                                X. Wang et al. 
sensory experience. Of course, it should be noted that the 
content of the work is also crucial, and VR effect will be 
less impressive if the audience cannot resonate with the 
plot.In the case of boring plot, even the best media will 
not bring the customer's love.Multimedia images and 
artwork formats are different from traditional images.The 
traditional image is mainly to get the customer's sensory 
effect. Multimedia images, on the other hand, are more 
focused on the immersion of the audience. Meredith 
bricken may be using the VR edge detection calculation 
method, if VR customers are fully into the state, they will 
be able to control the process.This effect is characteristic 
of multimedia images, a special relationship between 
viewers and images, which is a new model for learning 
knowledge.For example, views can choose fish (or birds) 
in the virtual world as they wish.The ability to gain 
knowledge from the animal's point of view, as well as the 
ability to both experience the divine feeling of the animal. 
In multimedia images, the audience is able to switch 
perspectives according to their preferences.In order to 
guide the audience's vision, the director of multimedia 
images should use the latest technology of photography at 
the beginning of the work to make it unfold smoothly. The 
viewing of multimedia images is not a one-sided act; it is 
an interactive form of dialogue. The computer is able to 
make timely updates according to the preferences of the 
audience, providing works corresponding to the needs of 
the audience. Of course, this interaction can also distract 
the audience's concentration [10-11].The more varied 
images of VR can especially disturb the audience's 
concentration and therefore can lead to a larger number of 
audiences losing interest in the director's shots.It is the 
result of intoxicated virtual world instead of senseless 
stories story. In the process of producing multimedia 
images, in order to overcome this series of drawbacks, the 
director uses long shots for correct stage deviation; stereo 
backgrounds, enhanced lighting and special effects in the 
form of attracting the audience's focus and making the 
director's pre-determined work unfold through 
perspective and creativity. When shooting a traditional 
film, the director's idea plus the camera's vision shoots the 
work through multiple angles, but the change of angle 
prevents the audience from getting into the plot of the 
work as soon as possible.Hence, the director completes 
the production through special editing and 
uniqueshooting angles.Multimedia and traditional images 
are different in that they are not influenced by the viewer's 
vision.VR can enter directly into the role of the 
work.From the above, we can see that the shooting of 
multimedia images is also different from the traditional 
video shooting.By using new technology, the work can be 
better displayed.It is also able to be adjusted according to 
the audience's experience in time [12-13].The accuracy 
and speed of VR edge detection calculation method for 
multimedia vision is now the primary issue.The accuracy 
of the calculation can be improved by the multimedia 
image edge detection calculation method, reducing the 
corresponding calculation time, and using the mean value 
method.The premise of retaining its original nature is 
applied to realistic situations by using virtual edge 
detection. Table 1 depicts the summary of related studies. 
Table 1: Summary of related works 
Ref. no Methods Findings Drawbacks 
[27]  Bluetooth Low 
Energy (BLE) 
The validation of the experimental 
system demonstrated that the fuzzy 
controller attains exceptional accuracy 
in detecting and avoiding obstacles, 
while also making directional decisions 
with minimal computational time. 
Despite its energy efficiency and low 
power consumption, Bluetooth Low 
Energy (BLE) faces certain 
drawbacks, such as a comparatively 
lower data transfer rate than classic 
Bluetooth. This limitation can impact 
the performance of applications that 
demand high data throughput. 
[28] Human machine 
interface (HMI) 
The smart glove, equipped with 
PDMS-CB strain sensors, effectively 
controls the remote motion of the robot 
finger, as demonstrated in experiments. 
 
A significant obstacle arises from the 
possibility of information overload, 
wherein an abundance of data and 
intricate interfaces has the potential 
to inundate users, resulting in both 
confusion and errors. 
[29] Level Of Robot 
Autonomy 
(LORA) 
While teleoperating systems have 
achieved advanced technical maturity, 
collaborative assisting and autonomous 
systems are currently in the research 
phase, emphasizing advancements in 
ultrasound image processing and force 
adaptation strategies. 
A notable drawback is its inclination 
to oversimplify the intricate nature of 
autonomous systems. LORA utilizes 
distinct levels, spanning from 
teleoperation to full autonomy, 
potentially failing to adequately 
grasp the nuanced capabilities 

Edge Detection and Simulation Analysis of Multimedia Images…                      Informatica 48 (2024) 97–110   99                                                                      
exhibited by robots. 
[30] Robot Operating 
System (ROS) 
The Robotics-Academy has proven 
effective in numerous undergraduate 
and master's degree engineering 
courses, as well as in a pilot course 
designed for pre-university students. 
ROS does not possess real-time 
capabilities, rendering it less ideal for 
tasks demanding accurate timing and 
low-latency control, such as specific 
industrial automation applications. 
Furthermore, a limitation arises from 
ROS's initial design geared towards 
research and prototyping, potentially 
hindering its seamless integration 
into extensive production 
deployments without additional 
optimization endeavors. 
 
2   Intelligent monitoring robot 
Multimedia images encompass digital materials that 
blend various forms of media components, including 
images, graphics, and occasionally audio or video, with 
the purpose of conveying information or crafting a 
visually captivating encounter. Within the realm of 
multimedia, the significance of images extends to their 
pivotal role in improving communication, storytelling, 
and the overall user experience. Their versatility is 
evident across a spectrum of applications such as 
websites, presentations, educational materials, 
entertainment, and beyond. Multimedia images exhibit a 
broad range of applications and are constantly adapting to 
technological advancements. As technology advances, the 
incorporation of multimedia components becomes 
progressively refined, delivering heightened levels of 
immersion and engagement for users. Intelligent 
monitoring robot vision imaging is used as the eyes of 
workers in the process of processing and manufacturing, 
which is an essential signal basis for intelligent 
monitoring robot to carry out the trajectory movement.It 
mainly converts the collected features such as color and 
shape of the workpiece into digital signals and transmits 
them to the general control system of the intelligent 
monitoring robot so as to control the operation of the 
lower machine [14]. Hence, all features required in the 
processing of the workpiece should be acquired during 
the signal acquisition process to ensure that the image is 
fully represented.The two key elements that can influence 
the effectiveness of vision imaging of intelligent 
monitoring robots are image acquisition and image 
processing. In addition to hardware facilities such as 
vision sensors, external light sources also have a 
significant influence on the stability of image 
acquisition.Therefore, a lighting system isoften added to 
the vision system of intelligent monitoring robots during 
the image acquisition.  
The hand-eye calibration method is defined mainly in 
accordance with the roles that the camera and the 
end-effector of the intelligent monitoring robot actuate, 
and the coordinate system of the camera is connected to 
the coordinate system of the intelligent monitoring robot 
to reflect its relative position relationship [15].According 
to the camera fixation method, there are two types, that is, 
Eye-in-Hand and Eye-to-Hand, which are shown in 
Figure 1 and Figure 2 below, respectively. 
 
Figure 1: Eye-to-Hand hand and eye detection system 
 
Figure 2: Eye-in-Hand hand and eye detection system 
In the depicted system shown in Figure 1, the camera 
remains stationary, while the robot tool coordinate system 
adapts to the trajectory's motion. Only the robot base 
coordinate system remains constant. Consequently, the 
camera coordinate system and the robot base coordinate 
system exhibit relative stability, allowing the 
determination of their positional relationship. The process 

100   Informatica 48 (2024) 97–110                                                                X. Wang et 
al. 
begins with determining the relative position 
characteristics of the workpiece captured by the camera 
once the robot vision system is in place. Subsequently, 
the position characteristics of the work piece are 
identified within the base coordinate system of the 
intelligent monitoring robot. Finally, the relative position 
relationship between the camera coordinate system and 
the base coordinate system of the intelligent monitoring 
robot is obtained through matrix transformation [16-17]. 
In the depicted system shown in Figure 2, the camera 
remains active as the robot base coordinate system 
remains stationary. In this scenario, the camera coordinate 
system cannot be stabilized in relation to the robot base 
coordinate system. Instead, the camera is affixed to the 
robot end-effector, and its coordinate system remains 
fixed in relation to the robot end-effector coordinate 
system. To achieve vision localization in the intelligent 
monitoring robot, the relative positional relationship 
between the camera coordinate system and the robot base 
coordinate system can be determined by acquiring the 
positional characteristics of the workpiece in both the 
camera coordinate system and the intelligent monitoring 
robot end-effector coordinate system. The camera 
parameter change matrix and the robot end position 
change matrix can be used to obtain the hand-eye 
relationship matrix.The two hand-eye calibration systems 
are compared, and the results suggest that the calibration 
based on the Eye-in-Hand hand and eye system is 
evidently more complicated, which is one of the popular 
topics in the research of vision system for intelligent 
monitoring robots at present. 
3  Multimedia image edge detection 
The multimedia image peripheral test module is designed 
with a focus on efficiently identifying and handling test 
objects. It mandates that the intelligent robot engaged in 
image discrimination processing initiates training on the 
current object beforehand, acquiring characteristic data of 
the target object to establish a useful reflection table. 
During system operation, the current work object is 
consolidated, and the extraction of characteristic values is 
activated. Subsequently, these values are compared with 
those in the reflective table, culminating in the successful 
identification of the target object. The specific design 
steps for the multimedia image peripheral test software 
are illustrated in Figure 3 below. 
 
 
 
Figure 3: Software design flow of the multimedia image 
edge detection module  
From Figure 3 above, it can be observed that the 
intelligent robot in the multimedia image around the test 
process to start the multimedia image in advance of the 
processing operation. In this way, the maximum possible 
simplification of multimedia image information, while 
the valuable information and detectability can be 
enhanced.Multimedia image cutting is atechnique and 
process of dividing multimedia images into a certain 
number of specific ranges with particular properties and 
referring to the target of interest.Multimedia image 
feature value acquisition is to identify the most efficient 
feature from a large number of featuresand to set up a 
feature reflection table for matching to the target object 
accordingly.The multimedia image matching is carried 
out based on the comparison of the acquired 
characteristic amount and the characteristic reflection 
table, which is used to achieve the purpose of identifying 
the target object. 
The feedback on the digital signals are given from the 
vision system of the intelligent monitoring robot to the 
general control system of the intelligent monitoring robot, 
which is required to collect information on the 
characteristics of the workpiece. When the camera 
acquires feature information of the workpiece, a large 
amount of redundant information is generated due to the 
external environment, which requires the camera to 
extract the required feature information from the huge 
amount of acquired information in a certain way. In this 
paper, the workpiece feature extraction is analyzed by 
two methods, that is, the threshold segmentation method 
and the edge detection method. 
 
 
 
 

Edge Detection and Simulation Analysis of Multimedia Images…                      Informatica 48 (2024) 97–110   101                                                                      
When the feature information of the workpiece is extracted by 
threshold segmentation, there are two types of threshold 
values, that is, fixed thresholds and automatic thresholds. The 
threshold value set by fixed thresholding is fixed. In the case 
of a changing external environment, the fixed threshold will 
not be changed along with the change of the environment, 
which can severely influence the accuracy of the captured 
image, as shown in Figure 4 below[18-19]. 
 
Figure 4: Acquired image of the workpiece in the fixed 
threshold case 
Based on the automatic threshold value, the algorithm can make 
adjustment automatically in response to the changes in the 
external image pixels and automatically select a suitable 
threshold value based on the gray point distribution of image 
pixels. This method is suitable for Eye-in-Hand hand-eye 
systems, in which the graphics captured by the camera change 
continuously with the robot motion and the graphics grayscale 
value. The details are shown in Figure 5 as the following. 
 
Figure 5: Adaptive threshold value 
After the threshold segmentation is carried out to obtain 
the feature information of the artifact, edge detection is 
often used to obtain acquire accurate graphic information. 
In addition, the methods such as Canny operator are 
applied to analyze and calculate the edge detection results 
and achieve the effect of accurate processing of the 
image[20]. 
In the calculation process, the global gradient is first 
calculated by smoothing the image based on Gaussian 
filtering, removing the pixel points where the local 
gradient is not the highest, keeping the pixel points with 
the largest local gradient, eliminating the false edge 
segments generated due to noise or color changes, and 
using a high threshold value and a low threshold value to 
distinguish the edge pixel points. In general situations, the 
high threshold value is about twice the low threshold 
value. Through suppressing the points smaller than the 
low threshold value, the effect of edge detection can be 
achieved. The details are shown in Figure 6 and Figure 7 
as the following. Finally, the workpiece edge pixel points, 
that is, the contours of the workpiece image, are extracted. 
 
Figure 6: Edge detection results 
 
Figure 7: Contour detection results 
3.1 Intelligent monitoring robot algorithm 
In accordance with the intelligent monitoring robot 
control algorithm, it is mainly implemented for the 
multimedia image edge detection visualization during the 
problem addressing process. For the purpose of resolving 
this problem thoroughly, it is necessary to optimize the 
parameters in the multimedia image recognition model by 
using the intelligent monitoring robot control algorithm in 
combination with the iterative strategy in this paper. The 
iterative value refers to the features of the intelligent 
monitoring robot control algorithm in using the intelligent 

102   Informatica 48 (2024) 97–110                                                                X. Wang et 
al. 
monitoring robot control algorithm in combination with 
the multimedia image recognition model based on the 
conventional vector, which is different in the sharing of 
parameters, while the intelligent monitoring robot control 
algorithm can enable the sharing between the local 
connection and the data multimedia image edge 
detection. 
Different from the traditional approach, in the intelligent 
monitoring robot control algorithm, the initial features of 
input signals in the data acquiredare used based on layer 
visualization. The difference in the mean amplitude of 
time domain features, the short-term energy, etc., are also 
applied. The initial input features of the multimedia 
image recognition model proposed in this paper apply the 
short-term energy in this paper [21-22]. Thus, the 
multimedia image recognition multimedia image edge 
detection signal can be expressed by using x(n), and the 
energy balance within the short time period can be 
expressed as the following: 
( ) ( ) ( ) ( ) ( ) ( )
2
22
n
mm
E x n w n m x m h n m x m h n

= −  = − 
= − = − =  


        
(1) 
However, the high dimensionality of the time-domain 
features under initialization can lead to a huge amount of 
interference and noise. In this way, it is required that the 
input visualization signal should to be processed for 
dimensionality reduction.At this point, the intelligent 
monitoring robot control algorithm is used to count the 
investigated multivariate and analyze the multivariate 
structure through the analysis of primary components. 
Dimensionality reduction is performed to process the data 
on multimedia images to visual the multimedia image 
edge detected in the input data. The output data are 
processed by the intelligent monitoring robot control 
algorithm based on deep learning. 
In multimedia image edge detection, feature vectors are 
observed at various positions during the visualization of 
the network, which showsthat the network can have 
different visualization capabilities. The level of feature 
vector is used to denote this property. During multimedia 
image recognition, if feature vector (TE) is observed n 
times, as shown in the equation (2) below, then 
k
T is the 
level of the kth occurrence of the feature vector T in the 
feature vector level score during the multimedia image 
recognition. 
( )
ln
1
1
2
k
n
T
k
score T
=
=

  (2) 
The feature vector weight can be expressed as the 
following 
( ) ( )
k
w T score T idf =  (3) 
When the similarity in weights between query Q and 
feature vector phrase D is calculated, the ratio of QD 
cross-section weight to total QD weight is used due to the 
large computational workload and high time overhead of 
the regular VSM similarity. 
( )
( ) ( )
( )
( ) ( )
1
11
,
QD
qk dk
k
QD
qk dk
kk
w T w T
sim Q D
w T w T
=
==
+
=

+





  (4) 
With regard to the vector space multimedia image 
recognition model, the conventional method for 
calculating the visual vector space for multimedia image 
edge detection is to calculate the cosine similarity 
between the vectors[23-24]. Thus, the visual vector space 
for multimedia image edge detection of user u and 
multimedia image d can be defined as the following 
( ) ,
ud
Sim u d
ud

=

 (5) 
The region distance metric is an essential benchmark for 
performing region integration, and the distance 
measurement benchmark can directly determine the 
results of region combination and the final image edge 
detection results.In general, the condition required to 
merge two regions is that the two regions are spatially 
adjacent and similar to each other in color and that no 
significant edges appear in their proximity. The color 
distance and edge distance between regions are specified 
accordingly. 
The color distance can be obtained as the following 
2
ij
c
ij i j
ij
rr
D
rr
 =−
+
(6) 
The edge distance can be obtained as he following 
( ) ( )
e
ij
D Ave i Ave j =−  (7) 

Edge Detection and Simulation Analysis of Multimedia Images…                      Informatica 48 (2024) 97–110   103                                                                      
Through the inverter nodes, and finally acting the actuator 
on everything in that environment shows the 
characteristics of autonomy, mutuality, purpose, sociality, 
cooperation, sustainability, adaptability and distribution. If 
the system needs to increase the number of remote 
monitoring nodes, the intelligent monitoring robot 
algorithm of this monitoring node need only to be adjusted 
to a communication mode suitable for this transformer less 
single-phase PV grid-connected inverter. Then, it is 
sufficient adding the information-behavior graph of this 
node. When the communication mode is changed in this 
way, the intelligent monitoring robot algorithm of the 
monitoring node is adjusted to fit the communication 
mode of this new electrified railroad. If a client queries a 
video image, the USG first confirms the client's ID and 
permission, connects to the video control center via the 
intelligent monitoring robot algorithm, and the video 
control center analyzes the letter based on the letter/action 
diagram after receiving the user message. Then, the 
corresponding remote monitoring node video grouping is 
forwarded to the client, confirmed to the client, or the 
corresponding alarm video data is found in the video 
database and returned to the client, and other actions. 
Single-phase PV grid-connected base-wave voltage 
s
v

 
interacts with the base-wave overrun current of the LC 
branch in the same phase transfers active power between 
the two converters, causing a change in the mean value of 
the DC capacitor voltage. Ignoring the converter loss, 
according to the law of conservation of energy, then 
( )
1
2
sd cd sq cq dc dc dc o
v i v i v Cv v i + = +     (8) 
Where, C is the DC bus capacitance. Orientate 0
sq
v =
based on the d-axis, then 
1
2
sd cd dc dc dc o
v i v Cv v i =+              (9) 
Let 
sd sd sd
cd cd cd
dc dc dc
o o o
v V v
i I i
v V v
i I i
= +  

= + 


= + 


= + 

                 (10) 
Here, the first term on the rightof equation (9) above 
indicates a steady state equilibrium quantity and the 
second term is the variation quantity. 
In equation (10), let the irrelevant perturbation 0
sd
v =
0
o
i =,, and substitute it into equation (9), neglecting
dc dc
v C v  , to obtain 
11
22
sd cd sd cd dc dc dc o dc o
V I V i V C v BV I v I +  =  + +                
(11) 
Since the two converters have to ensure the energy balance, 
they must satisfy 
1
2
sd cd dc o
V I V I =                    (12) 
Then equation (13) is further simplified as 
( )
( )
1
2
sd
dc
cd dc o
V
Vs
I s CV s I

=
+
            (13) 
Let
dc sd
VV  = ,  is the adjustable coefficient,
o dc o
I V R = Formula (13) can be reduced to 
( )
( )
1
21
dc o
cd o
Vs R
I s R Cs 

=
+
            (14) 
Intelligent monitoring robot clustering algorithm is 
currently the most popular and one of the most widely 
used cluster analysis methods.The intelligent monitoring 
robot algorithm has the features of intuitive and easy to 
implement image edge detection at high speed, but the 
biggest problem is to pre-determine the number of clusters. 
In addition, the selection of the initial cluster centers has a 
significant impact on the classification results. On the 
other hand, image edge detection using the clustering 
algorithm is sensitive to noise due to the use of only the 
color information of the pixels without using that spatial 
information, which can cause image edge detection. 

104   Informatica 48 (2024) 97–110                                                                X. Wang et 
al. 
3.2 Improved algorithms for multimedia 
image edge detection 
When calculated according tothe correlation function, 
individual feature vector of the target in the image subject 
to multimedia image edge detection can affect the data on 
edge detection substantially. Errors are prone to occur if 
target individual stripes have few features, making it 
harder to reconstruct the image and enhance the specific 
stripes of those reconstructed. Hence, the features of 
original image patterns and the accuracy of results can be 
improved by superimposing high-frequency data from 
raw imageson the selected image. 
After denoising,
x
I becomes the image grayscale sequence 
in multimedia image edge detection, where x = P. N, In the 
case of obtaining the target individual edge region feature 
distribution, the grayscale conversion equation 
1
0 1 1
, , , 2
Q
i
c Q i
binary
i
Dce
WW
j
ix
j
S S S S S
SI
−
−


 = = 
 

=


                 
(15) 
Here, Q representsthe grayscale next to the target region, 
W represents the conversion step, and the noise removal 
function is 
( ) ( ) ( )
( ) ( ) ( ) ( )
1 , 1,2, ,
, 1,2, ,
ii
x
ii
x
x k Q k w k k m
z k H k k v k i m
+ = + =
= + =
(16) 
where ( )
i
wk and ( )
i
vk } represent pixel noise in the 
edge region of target individuals, ( )
i
Qk and ( )
i
Hk
are distributed in a balanced manner (mean 0, variance 
( )
i
Sk to buildthe fuzzy set  
ik
uu = andbreak down 
the stripe featuresto obtain the output of denoising as 
follows 
( ) ( ) ( ) ( ) ( )
( ) ( ) ( )
11
1
;;
NN
N N N
GSM i i i i i i i i
ii
N
i i i i i
i
I I C D s I C D s h D s h D C s
h g C V s h V
==
=
= = = −
= + −


       
(17) 
Considering the differences between edge detection image 
regions, information on spatial ripple characteristics is 
described by 
( )
( ) ,,
n
u x y d , pixel gray intervals are built 
in gradient to obtain an iterative equation for denoising of 
images: 
( )
( )
( )
( )
( )
( )
1
1
, , ,
n n n
u x y u x y u x y 
+
=+ (18) 
( )
( )
( )
( )
( )
( )
1
, , , ,
n n n
xt
u x y M u x y N u x y d =  +  (19) 
Here 1,2, , nT = is used to denote the number of 
iterative steps. The interference noise removal method 
reduces its influence on the image, and it completes the 
improvement of the intelligent monitoring robot 
calculation method. 
In the process of determining the directional coordinates
( )
12
, r n n , if the global search is applied, it will lead to 
the error of the peripheral detection. Therefore, the search 
scope should be decreased by rebuilding the 
interconnectedness between levels and plan to reduce the 
occurrence of the error of the peripheral detection. 
According to the provisions of the horizontal two-fold 
limit, if the corrected longitudinal deviation continues to 0, 
then the longitudinal search range can be reduced. In 
CTF's 3D peripheral detection composition, according to 
the presumption of arbitrary reconstruction level of 
connectivity, it can reduce the lateral search area [25]. 
The maximum value of the low-pass filtering process is 
calculated again for the increase in ( )
12
, r n n . The 
maximum value range can be reduced as shown in 
Equation (20) 
( )
1 1 2 2
12
1, ,
,
0,
k U k U
h k k
otherwise
  

=



      (20) 
If the corresponding offset interval between horizontal and 
vertical is very small, then the intelligent monitoring robot 
algorithm cannot guarantee the accuracy of offset data, 
specifically in the case of a large number of values, and 
the correction function of the CTF architecture cannot be 
used properly. The restricted area is too large and does not 
have the ability to reduce the error reconstructed. 
Therefore, the optimal search range is determined based 
on U1 and U2. 
(2) Misidentification and peak relocation 
 

Edge Detection and Simulation Analysis of Multimedia Images…                      Informatica 48 (2024) 97–110   105                                                                      
In the initial calculation method, after the reconstruction, 
the wrong reconstruction can be evaluated and modified 
only once, but in practice there is a process of intermediate 
reconstruction, which leads to the extension of the 
reconstruction error to the next resolution. At the same 
time, the evaluation and correction criteria are fixed 
critical values 
th
 and maximum values of α. However, 
the comparison results use a more accurate dynamic 
device to determine the critical values
th
 , because the 
full maximum value of each resolution varies greatly [26]. 
The minimum feature value r(i,j) of the image 
acquiredbased on the intelligent monitoring robot control 
algorithm is the image edge feature to be calculated, 
followed by calculating the difference between feature r 
and feature(i,j) in coordinates. Thus, the optimal 
visualized image can be obtained as required. Upon the 
calculation based on the intelligent monitoring robot 
control algorithm corresponding to the elements of the 
two feature arrays, it is required to determine whether it 
exceeds the minimum at present. If so, the intelligent 
monitoring robot control algorithm is ended accordingly 
during the calculation process. 
The mapping is generalized to obtain the corresponding 
generalized mapping equation as the following: 
0
0
mod
n
n
n
xx ab
N
yy cd
    
=
    
    
(21) 
From the equation (1) above, it can be known that there is 
an immobile point (0, 0) in this mapping. In other words, 
the point (0,0) will remain unchanged even after n 
iterations of the mapping. Subsequently, the value of the 
coordinate point is changed to {1, 2, N} so as to avoid the 
immobility point. In accordance with {1, 2,… N}, the 
mapping equation is converted to a format that contains 
two independent parameters as follows. 
0
0
1
mod 1
1
n
n
n
xx q
N
yy q pq
    
=+
    
+
    
(22) 
The equation (22) above expresses the result by taking 1 
as the transformed coordinates ( ) ,
nn
xy after n 
repetitive operations based on the model N from the 
initial value point ( )
00
, xy . 
4  Practical cases and data analysis 
The DCNN networks run in Intel on a Core i7 PC with 40 
GB RAM. Our proposed model was trained on an 
NVIDIA GTX 1070 GPU and implemented by Python. 
The test for the image edge detection performance can be 
assessed mainly based on the precise location of the 
image edge, the continuity of the image edge, and the 
edge width. In this paper, a king of assessment method 
for image edge performance is adopted, which is defined 
as the following: 
( )
1
1
1
max ,
1
D
I
DL
i i
p
II
I 
=
=
+

         (23) 
The CT image of the human head is selected as the target 
for simulation evaluation, and the effectiveness of the 
image edge detection method put forward in this paper is 
tested accordingly. The details are shown in Figure 8 
below. At the same time, the Canny algorithm and the 
Sobel algorithm are used for comparison and analysis. 
 
 
Figure 8: Selected CT image of the human brain 
The results of the test obtained based on the different 
methods are shown in Figure 9 as the following.In 
accordance with the analysis of the test results in Figure 9, 
it can be observed that the method adopted in this paper is 
more ideal as compared to the image edge test method; 
the image edge continuity thus obtained is high, and the 
accuracy of detection is also the highest. Excessive 
details of the image areacquired from the test based on 
the Canny algorithm. In addition, some of the edge details 
are not detected, and the image has been distorted. The 
incorrect detection and omission in the image edge test 
obtained based on the Sobel algorithm are 

106   Informatica 48 (2024) 97–110                                                                X. Wang et 
al. 
relativelysevere, which fails to achieve the expected 
results. 
 
Figure 9. Effect of image edge detection based on 
different methods 
As shown in Figure 8 above, it can be observed thatif 
there is interference of different degrees of external noise, 
followed by the application of different methods to obtain 
the edges of the image, the analysis results of the P value 
for the evaluation index can be obtained. The details are 
shown in Figure 10. In accordance with the test results 
illustrated in Figure 10, relatively good curve fitting can 
be achieved by using the proposed method, which can 
improve the detection accuracy of image 
edgeseffectively.The comparative analysis suggests that 
the proposed method is superior interms of noise 
resistance and high detection accuracy in the process of 
image edge detection. 
 
Figure 10: Performance index of image edge detection 
based on different methods 
The method in this paper is applied to the multimedia 
image design department of an enterprise, mainly users 
evaluate the works of multimedia image designers in this 
department to further grasp the shortcomings of the works 
and implement improvements. The adoption rate results of 
different types of multimedia image works tested after 
using this method in this department are shown in Figure 
11.  
By analyzing Figure 11, it can be seen that after applying 
the method of this paper, the lowest adoption rate of 
multimedia image design in this department is 0.98, and 
many multimedia images works in this department are 
adopted. The figure indicates that the method has 
outstanding performance and can be used as one of the 
application methods for multimedia image companies to 
improve the quality of multimedia images. 
 
Fig. 11: Test results of intelligent monitoring robot 
algorithm. 
4.1 Peak signal-to-noise ratio (PSNR) 
 
It's a statistic that's frequently used to assess the manner 
in which compressed or rebuilt photos and movies are 
doing. The ratio of a signal's maximal potential strength 
to the amount of corrupting noise that degrades the 
representational fidelity is called PSNR. Table 2 and 
figure 12 depicts the outcomes of PSNR. 
 
 
Table 2: Comparison of PSNR 
 
Methods MSE 
V-bpp Edge-XOR [31] 0.288 
Improved Hash [31] 0.409 
Fractional Fourier Transform [31] 17.58 
Median Filter [31] 24.66 
Ant Colony Optimization [31] 8.233 
Proposed 0.125 
 
 

Edge Detection and Simulation Analysis of Multimedia Images…                      Informatica 48 (2024) 97–110   107                                                                      
 
Fig. 12: Analysis of PSNR. 
4.2 Mean squared error (MSE) 
 
The average of the squared variance of the actual and 
projected values is measured using the mean squared 
error (MSE) metric in analytics and machine learning. It 
is computed as an average of the squared discrepancies 
among the anticipated and actual results, and is frequently 
used in regression analyses. Table 3 and figure 13 depict 
the outcomes of MSE. 
 
Table 3: Comparison of MSE 
 
Methods PSNR 
V-bpp Edge-XOR [31] 53.532 
Improved Hash [31] 47.559 
Fractional Fourier Transform [31] 25.786 
Median Filter [31] 18.85 
Ant Colony Optimization [31] 8.975 
Proposed 62.872 
 
 
 
Fig. 13: Analysis of MSE. 
4.3 Discussion 
Bluetooth Low Energy (BLE) [27] has various limitations, 
such as a slower data transmission rate than traditional 
Bluetooth, even though it is acknowledged for its energy 
economy and low power consumption. One of the key 
challenges in the field of Human Machine Interface (HMI) 
[28] is the possibility of information overload, which 
occurs when users are inundated with too much data and 
complex interfaces, which can result in mistakes and 
confusion. Moreover, ROS's lack of real-time capabilities 
makes it less appropriate for applications requiring low 
latency control and accurate timing. One important 
drawback of Level of Robot Autonomy (LORA) [29]. Its 
tendency to oversimplify the complexity of advanced 
autonomous systems is a notable aspect. Nevertheless, our 
proposed approach adeptly navigates and effectively 
mitigates these challenges, ensuring a comprehensive and 
satisfactory resolution to the aforementioned difficulties. 
5  Conclusion 
In the process of reconstruction, data information 
transmission, and sending and receiving of digital 
multimedia images, due to the limitations of multimedia 
image restoration device quality and multimedia image 
edge detection methods, different types of edge feature 
interference of multimedia images will emerge. In VR, the 
audience can adjust the line of sight and enquire the 
information onthe environment, which is impossible 
based on traditional images.This paper adopts the 
matching registration to lock the peak range of the image, 
calculates the threshold and optimizes the model, 
evaluates and corrects the wrong data, and calculate the 
max, min, and depth.After improvement, the noise during 
transition can be reduced, and the calculation is 
performed at a higher speed. It has a better feeling for 
multimedia images and maintains both robustness and 
feasibility. Based on the weight of the intelligent 
monitoring robot, the relationship between each pixel of 
the multimedia image and the size of the pixel to be edge 
detected can be calculated, and the correlation between 
multimedia image pixel and local data can be used 
appropriately. To a certain extent, it can effectively solve 
the processing of multimedia image edge features and 
multimedia image edge details.Finally, the experimental 
research shows that the intelligent monitoring robot used 
in this paper has a higher PSNR value than the traditional 
multimedia image edge detection algorithm, and the 
iterative adaptive method can effectively process the edge 
details of the multimedia image, and has a better 
multimedia image processing effect. 

108   Informatica 48 (2024) 97–110                                                                X. Wang et 
al. 
Limitations and future study 
The capacity of an appliance to retain its functionality 
and efficacy under varied situations and in the face of 
numerous obstacles is referred to as robustness in many 
domains, such as machine learning, programming, or 
safety. The resolution of the monitoring robot's cameras 
can constrain the quality of multimedia images, with 
higher-resolution cameras typically being more costly 
and resource-intensive. The swift analysis and 
interpretation of images by monitoring robots may be 
impeded due to constraints in processing capabilities, as 
real-time processing of multimedia data demands 
significant computational power. Enhancing the robot's 
perception capabilities, especially in demanding 
environmental conditions, is achievable by incorporating 
a variety of sensors like infrared, lidar, radar, and 
cameras. It is crucial to establish ethical guidelines and 
legal frameworks for deploying monitoring robots, with a 
focus on prioritizing transparency, accountability, and the 
protection of privacy rights in upcoming research 
endeavours. 
Data aailability 
The data used to support the findings of this study are 
included within the article. 
 
Conflicts of Interest 
The authors declare no conflicts of interest. 
 
Funding statement 
This research study is sponsored by these projects: 
project one: The Humanities and Social Sciences 
Research Projects of the Ministry of Education, the name 
of the project is Research on university learning behavior 
model based on the fusion of spatio-temporal trajectory 
data and multi-source data, the project number is: 
19C10481026. Project two: Key research and 
development and promotion projects in Henan Province, 
the name of the project is Research and development of 
medical endoscope system based on narrow band imaging 
and intelligent image fusion, the project number is: 
192102310457. Thank these projects for supporting this 
article! 
 
 
 
 
References 
[1] Yue, J. J., & Chen, G. (2020). Competence of 
pharmacy mentors: a survey of the perceptions of 
pharmacy postgraduates and their mentors. BMC 
Medical Education, 44(2), 113-117. 
[2] Li T Y. (2021). The construct of english competence 
and test design for non-english major postgraduates. 
English Language Teaching, 23(5), 531-55. 
[3] Moskal, M., &Schweisfurth, M. (2018). Learning, 
using and exchanging global competence in the 
context of international postgraduate mobility. 
Globalisation, 34(7), 1104-1109. 
[4] Dong, Q. (2020). Analysis on the ways to improve 
thesis writing ability of postgraduates in design 
direction of  
[5] Xiao, Y., Wu, X. H., Huang, Y. H., & Zhu, S. Y. 
(2021). Cultivation of compound ability of 
postgraduates with medical professional degree: the 
importance of double tutor system. Postgraduate 
Medical Journal, postgradmedj-2021-139779. 
[6] Li T Y. (2021). The construct of english competence 
and test design for non-english major postgraduates. 
English Language Teaching, 57, 237-248. 
[7] Liang, L., Li-Ye, H. E., & Chen, B. J. (2020). 
Discussion on the cultivation of scientific research 
ability of postgraduates majoring in medicine. 
DEStech Transactions on Social Science Education 
and Human Science(aems), 38(1), 55-61. 
[8] Feng, L. . (2016). Study on practical innovation 
ability cultivation of management postgraduates 
based on qfd. DEStech Transactions on Social 
Science Education and Human Scienc, 39(6), 
1006-1011.. 
[9] Taghizadeh, N. (2020). Research on the practical 
ability of postgraduates majoring in marxist 
theory—based on a survey of universities in 
chongqing. Advances in Social Sciences, 09(5), 
641-650.. 
[10] Rezaie, A. A. , &Habiboghli, A. . (2017). Detection 
of lung nodules on medical images by the use of 
fractal segmentation. International Journal of 
Interactive Multimedia & Artificial Intelligence, 
4(5), 15-19.. 
[11] Sun, H. L. ,  Peng, X. , &  University, Y. . (2018). 
On practical ability improvement-oriented 
cultivating mode of professional degree 
postgraduates. Heilongjiang Researches on Higher 
Education, 2021(4),  
[12] Han, X. ,  Singh, B. ,  Morariu, V. , &  Davis, L. 
S. . (2015). Vrfp: on-the-fly video retrieval using 
web images and fast fisher vector products. IEEE 
Transactions on Multimedia, 1-1. 

Edge Detection and Simulation Analysis of Multimedia Images…                      Informatica 48 (2024) 97–110   109                                                                      
[13] Su, G. M. ,  Zhang, C. ,  H  Wang,  Chen, M. , 
&Lienhart, R. . (2019). Pushing the boundary of 
multimedia  
[14] Li, &Xirong. (2017). Multimedia systems 
(accepted) (will be inserted by the editor) tag 
relevance fusion for social image retrieval. 
Multimedia Systems, 23(1), 29-40. 
[15] Kim, W. ,  Lee, S. , &Bovik, A. C. . (2021). Vr 
sickness versus vr presence: a statistical prediction 
model. IEEE Transactions on Image Processing, 30, 
559-571. 
[16] Zhao, S. , Shmaliy, Y. S. , Ahn, C. K. , & Liu, F. . 
(2018). Adaptive-horizon iterative ufir filtering 
algorithm with applications. IEEE Transactions on 
Industrial Electronics, 65(8), 6393-6402. 
[17] Zheng, J. , Yang, P. , Chen, S. , Shen, G. , & Wang, 
W. . (2017). Iterative re-constrained group sparse 
face recognition with adaptive weights learning. 
IEEE Transactions on Image Processing, 26(5), 
2408-2423.. 
[18] Zeng, J. , Sui, X. , & Gao, H. . (2015). Adaptive 
image-registration-based nonuniformity correction 
algorithm with ghost artifacts eliminating for 
infrared focal plane arrays. IEEE Photonics Journal, 
7(5), 1-16.. 
[19] Ahmed, F. , & Das, S. . (2014). Removal of 
high-density salt-and-pepper noise in images with 
an iterative adaptive fuzzy filter using 
alpha-trimmed mean. IEEE Transactions on Fuzzy 
Systems, 22(5),  
[20] Chen, G. , Li, D. , & Zhang, J. . (2014). Iterative 
gradient projection algorithm for two-dimensional 
compressive sensing sparse image reconstruction. 
Signal Processing, 104, 15–26. 
[21] Hao, Z. , Xianqi, L. , Yunmei, C. , Jewook, P. , 
An-Ping, L. , & X.-G., Z. . (2017). Postprocessing 
algorithm for driving conventional scanning 
tunneling microscope at fast scan rates. Scanning, 
2017, 1-8. 
[22] Chen, Z. , Fu, Y. , Xiang, Y. , & Rong, R. . (2017). 
A novel iterative shrinkage algorithm for cs-mri via 
adaptive regularization. IEEE signal processing 
letters, 24(10), 1443-1447.. 
[23] Jiang, G. ,  Luo, M. , &  Bai, K. . (2019). Optical 
positioning technology of an assisted puncture robot 
based on binocular vision. International Journal of 
Imaging Systems & Technology, 29(2), 180-190. 
[24] Wang, T. , &  Guo, J. . (2018). Design and 
implementation of robot precise grasp based on 
image processing. Manufacturing Technology & 
Machine Tool, 24, 1012–1017. 
[25] Sheu, J. S. , &  Tsai, W. H. . (2017). 
Implementation of a following wheel robot featuring 
stereoscopic vision. Multimedia Tools and 
Applications, 76(23), 25161-25177. 
[26] Horssen, E. V. ,  Hooijdonk, J. V. ,  Antunes, D. , 
&Heemels, W. . (2019). Event- and deadline-driven 
control of a self-localizing robot with 
vision-induced delays. IEEE Transactions on 
Industrial Electronics, 9(5), 54-63. 
[27] Narayanan, K.L., Krishnan, R.S., Son, L.H., Tung, 
N.T., Julie, E.G., Robinson, Y.H., Kumar, R. and 
Gerogiannis, V.C., (2022). Fuzzy guided 
autonomous nursing robot through wireless beacon 
network. Multimedia tools and applications, 
pp.1-29. 
[28] Dong, W., Yang, L. and Fortino, G., (2020). 
Stretchable human machine interface based on smart 
glove embedded with PDMS-CB strain sensors. 
IEEE Sensors Journal, 20(14), pp.8073-8081. 
[29] von Haxthausen F, Böttger S, Wulff D, Hagenah J, 
García-Vázquez V, Ipsen S (2021). Medical robotics 
for ultrasound imaging: current systems and future 
trends. Current robotics reports.2:55-71. 
[30] Cañas, J.M., Perdices, E., García-Pérez, L. and 
Fernández-Conde, J., (2020). A ROS-based open 
tool for intelligent robotics education. Applied 
Sciences, 10(21), p.7419. 
[31] Park, K., Chae, M. and Cho, J.H., 2021. Image 
pre-processing method of machine learning for edge 
detection with image signal processor enhancement. 
Micromachines, 12(1), p.73.