Remote medical video region tamper detection system based on Wireless Sensor Network

INTRODUCTION: A new telemedicine video tamper detection system based on wireless sensor network is proposed and designed in this paper. OBJECTIVES: This work is proposed to improve the performance of telemedicine video communication and accurately detect the tamper area in telemedicine video. METHODS: The sensor nodes in the sensing layer are responsible for collecting telemedicine video information and transmitting the information to the data layer. The data layer completes the storage of information and transmits it to the processing layer. The detection module of the processing layer detects the tampered area of the telemedicine video through two parts: suspicious moving point calculation and tamper detection, and transmits the detection results to the application display layer for display. RESULTS: The experimental results show that the designed detection system can accurately detect the tampered area in the telemedicine video, and the packet loss rate is significantly reduced, and the maximum packet loss rate is no more than 1%. CONCLUSION: The proposed detection system for remote medical video based on wireless sensor network can better meet the requirements of region tamper detection.


Introduction
From the current technical level, the level of medical information in our country is still in the stage of exploration and research. Especially in clinical nursing, most hospitals still use manual operation, which leads to the lack of accurate collection and classification of complex clinical nursing information and the poor effect of clinical nursing [1]. In order to relieve the pressure of medical staff and improve the medical level according to information technology, it is an urgent problem to be solved. Today, with the rapid development of wireless sensor networks, smart medicine has begun to enter people's vision and become an indispensable part of the development of medical information [2]. Smart medicine is to monitor and transmit the physiological parameter information of patients in real time through computer technology combined with wireless sensor networks, so as to quickly and accurately solve the problems of patients' medical needs. Telemedicine is an important part of smart medicine. Because some experts are far away from patients and it is difficult to reach them in a short time, it can be carried out through telemedicine or remote surgery. Therefore, a large amount of video resources need to be transmitted in the process of telemedicine. In order to create medical accidents or make patients panic, some illegal elements will tamper with the video resources of telemedicine, resulting in greatly reduced medical effects. Therefore, it is necessary to detect the tampering in the video area of telemedicine, which can avoid the above problems and improve the effect of telemedicine.
Wireless sensor network (WSN) is a multi hop self-organizing network system composed of a large number of cheap micro sensor nodes deployed in the monitoring area through wireless communication. Its purpose is to collaboratively sense, collect and process the information of the perceived objects in the network coverage area, and send it to the observer. The three elements of wireless sensor networks are sensors, observers and sensing objects [3]. Sensor nodes are composed of power supply, sensing unit, embedded processor, memory, communication unit and software. The observer is the user of the sensor network, the receiver and the applicator of the perceptual information. The observer can be a person, a computer or other device. The sensing object is the monitoring object of interest to the observer, which can be humidity, temperature, light, pressure, etc. Wireless sensor networks integrate the logical information world with the objective physical world, changing the interaction between human and nature [4]. People can directly perceive the objective world through sensor networks, thus greatly expanding the functions of existing networks and human ability to understand the world. Wireless sensor networks usually include sensor nodes, transmit nodes and sink nodes. A large number of sensor nodes are randomly deployed in or near the monitoring area. Under the guidance of the cluster head node, a routing topology is established.
Then the sensor nodes collect and record the environment information of interest around [5], and transmit hop by hop along the previously established routing topology path.
During the transmission process, the data may be processed by multiple forwarding nodes, and transmitted to the sink node after a single hop or multi hop routing [6], the sink node transmits the data to the gateway node through wired mode for centralized processing.
With the continuous development of Internet technology, illegal users can modify images or videos through image or video editing software, resulting in a serious reduction in the authenticity and effectiveness of image or video resources [7]. In order to ensure the authenticity of the original video content, academia has proposed many active forensics technologies, such as digital watermarking technology. The authenticity is verified by embedding authentication information when recording video, but active forensics technology requires embedding authentication information when recording video, which is difficult to achieve in real scenes. Therefore, passive forensics technology based on the characteristics of video itself has greater development and application space. Video's inter-frame tampering detection technology is the main branch of passive forensics technology [8]. Remote medical video region tamper detection system based on Wireless Sensor Network mainly divided into two categories: methods based on the content discontinuity of tamper points and methods based on the periodic effect of secondary coding. There are too many images in the telemedicine video area, which is easy to be tampered with [9].
Lixiaohong and others proposed a mobile device image tamper detection method based on pruning compression CNN method [10]. Through the fusion of activation value and information entropy, the importance of CNN weighting can be effectively evaluated and the weighting with low importance can be cut off. Feedback adjustment is made according to the accuracy and pruning effect to control the balance of pruning compression. For pruning compression CNN, the corresponding convolution layer, pooling layer and adjustment layer are designed, which are analyzed and optimized from the perspective of layer and tampering mode respectively, and the tampering location is determined according to the correlation of image blocks. Gharbi et al.
Used unsupervised Bayesian method to detect remote sensing images [11], accurately collected the information of video images through hierarchical Bayesian model and Gibbs sampler, and applied change detection based on Bernoulli model to remove the noise of the object to be detected, so as to ensure the operation efficiency of the algorithm. Yan Pu et al. Applied the multi support region local brightness order method to the forgery and tampering detection of video images [12], extracted the affine invariant region as the support region by using the maximum stable extreme region (mser) algorithm, and obtained multiple support regions with different scales, resolutions and directions by using the non sampling contourlet transform. The liop descriptors with rotation invariance and monotone luminance invariance are extracted from each support region, and the bidirectional distance ratio method is used to realize feature initial matching. Spatial clustering is used to classify the matched features, then random sampling consistency (RANSAC) algorithm is used to estimate the geometric transformation parameters of each classification, and necessary post-processing operations are used to detect the forged and tampered areas. The above three methods can achieve image tamper detection, but they have the defects of poor communication performance and can not be applied to remote medical video tamper detection. Therefore, the area tampering detection system of telemedicine video based on wireless sensor network is studied in this paper. The overall design scheme of the system is: (1) The overall structure of the remote medical video region tamper detection system is designed. The system consists of five layers: perception layer, data layer, processing layer, network layer and application layer.
(2) The hardware of the system is designed according to the overall architecture of the system. The hardware includes wireless sensor network module and ZigBee chip.
The wireless sensor network is used to collect the information of telemedicine video area, and ZigBee chip is responsible for the transmission of telemedicine video.
(3) According to the calculation results of suspicious moving points in the telemedicine video area, the suspicious video sequence of the telemedicine video is grayed, and the SIFT algorithm is used to extract the feature points of the frame, so as to complete the detection of tampered areas.

Overall system structure
Telemedicine video's area tampering refers to that a key region of the video frame image is covered or replaced.
After image editing and repair, the tampering trace is difficult to be distinguished by the naked eye. If the tampering operation is a malicious forgery, it will have a very serious impact and consequences. Therefore, the research on the detection and location of area tampering in telemedicine video has important research value and application prospect. According to the architecture of wireless sensor network technology, the area tampering detection system of telemedicine video is divided into five layers: perception layer, data layer, processing layer, network layer and application layer. The overall structure of the system is shown in Figure 1.   Compared with other wireless transmission protocols [13], ZigBee specification has the application of wireless sensor networks that support star, tree and mesh network topologies, low power consumption, low cost and low data rate. In the specific development of the system, the beestack protocol stack of Freescale company is adopted, which uses direct sequence spread spectrum to divide the 2.4GHz frequency band into 16 non overlapping optional channels, so that the transmission of different channels will not interfere with each other. The disadvantage of ZigBee specification is that it adopts centralized management and requires gateway equipment.

(2) Data layer
The data layer includes five parts: monitoring database, standard physiological parameter database, expert database, nursing database and communication database. Each database in the data layer is used to store different types of data run by the system.

(3) Processing layer
The sensor device of the system transmits the collected data to the processing layer. The processing layer uses the video  dynamic storage and calculation of massive data, realize the scientific rationality of data processing [15][16], query the diversity and load balance of server data sources, and promote the development of "intelligent medical treatment" and hospital service utilities.

(5) Application layer
The application layer of the system adopts B/S architecture, the processing layer uses C# language to transmit the processing information to SQL Server for storage through the network layer, and the application layer uses ASP Net language to realize the dynamic display of data. The client can log in  Since the built-in processor of JN6201 is a 32-bit RISC core with a speed of 16MHz, it can be used as the CPU of the node, which saves the cost of using an independent CPU and simplifies the hardware structure. The data transmission interface adopts the UART port of JN6201. The master node connects with the PC through this port, uploads data and receives instructions from the upper computer software.
The slave node can connect with the RFID read-write module through this port as the interface for the RFID system to access the ZigBee network.
The onboard sensors on the slave node include general sensors and medical sensors, in which any digital sensor for testing vital signs can be selected as the medical sensor. We take the HKX-08A heart rate sensor module provided by Huake Electronic Research Institute as an example. It is a circuit module that integrates analog signal processing and digital processing technology to detect heart rate. The slave node can obtain its heart rate data through serial port.
In addition, in order to meet the processing needs of heart rate sensor, STC12C5A60S2AD microprocessor unit can be added. In this system, RFID is used to identify the identity of personnel to verify their legitimacy. RFID system includes reading and writing module and tag. The reading and writing module in this system adopts (iv) High wireless transmission power; (v) Four power management modes. The current consumption in standby mode is less than 0.3uA, and external interruption can wake up the system; (vi) Low voltage power supply is adopted (1.5V ~ 3.3V); (vii) VOA and LNC are integrated on chip; (viii) Four 8-bit ADC are integrated; (ix) Integrated with IEEE802.15.4 standard transceiver; (x) AES safety coprocessor is integrated; (xi) It has the functions of battery detection and temperature sensing detection.
As a monolithic ZigBee chip with integrated CPU, ZIC2410 provides a high-performance and low-cost RF transceiver scheme for ZigBee network, which complies Remote medical video region tamper detection system based on Wireless Sensor Network with ZigBee specification and IEEE802 15.4 standard, and the highly intensive integration simplifies functional modules, reduces power consumption, and reduces the cost of the whole system to a new level. Zic2410 chip integrates 8051F020 microprocessor, which can realize data processing and control of external I/O port; It has 96KB FLASH, which can store data; It integrates temperature sensor, wireless transceiver module and so on.

Design of system software
The video's tampering area detection module realizes the detection of tampering area in telemedicine video through two parts: the calculation of suspicious moving points in telemedicine video and the detection of tampering area in telemedicine video.

Calculation of suspicious moving points in telemedicine video
At present, the frame-by-frame tampering repair technology to delete a target in the video is mainly to fill the deleted area through some repair algorithms. The frame sequence of video is continuous, and the repair operation is often difficult to ensure the continuity and consistency between the unmodified frame and the tampered frame, so that the  Remote medical video region tamper detection system based on Wireless Sensor Network For a pixel ( ) , xy , when the gray value of the point changes greatly in the time domain and is not an edge information point, it is considered that the point may be tampered with, which is called a suspicious motion point, and the telemedicine video image sequence is a suspicious video sequence. According to the sequence results, the feature points of the frame can be extracted by SIFT algorithm to achieve regional tamper detection.

Detection of tampering in telemedicine video area
The suspicious sequence in the telemedicine video image sequence is detected through the above calculation, and the

Results
In order to verify the effectiveness of the designed area

Experimental data
Collect the original monitoring information of telemedicine equipment as experimental data. The original monitoring image collected by telemedicine equipment in the hospital is shown in Figure 4.

Experimental scheme and index
In order to fully verify the performance of the designed system, the tamper area location, energy suspicious degree and area tamper detection effect are taken as the indicators of system performance verification to verify the performance of the system in this paper. Taking the packet loss rate of the system operation as the comparison index, the system in this paper is compared with the pruning and compression CNN system proposed in reference [10] and the unsupervised Bayesian elimination proposed in reference [11].

Tampering area positioning
The system designed in this paper is used to detect the regional tampering of telemedicine video, and the detection results are shown in Figure 5.   The red border gray area in Figure 5 is the tampering area of telemedicine video detected by the system in this paper. Through the comparison results of Figure 4 and Figure 5, it can be seen that the system in this paper can effectively detect the tampering area of telemedicine video.
The area tampering detection results of telemedicine video is used to ensure the reliability of telemedicine video monitoring, and the accurate remote video monitoring results are used to improve the application effect of telemedicine video monitoring. This system can clearly detect the tampered area of telemedicine video, and can effectively locate the tampered area. When the tampering area is large, it still has effective detection effect.

Energy suspicion
The system in this paper is used to detect the tampered area

Regional tamper detection effect
The system in this paper is used to detect the tampered area in the telemedicine monitoring video, and the detection results of the tampered area in the telemedicine monitoring video within 60min are counted. The statistical structure is shown in Table 1.
It can be seen from the experimental results in Table 1 that the system can effectively detect the tampering area of telemedicine monitoring video, and the accurate detection  detection system. The system in this paper is used to detect the tampering area of telemedicine monitoring video, the packet loss rate and transmission delay of transmitting different types of medical data. The system in this paper is compared with the system in reference [10] and the system in reference [11]. The statistical results are shown in Figure   7 and Figure 8. As can be seen from the experimental results in Figure 7 and Figure 8, the system in this paper adopts wireless sensor network as the communication technology for tampering area detection of telemedicine monitoring video, and the communication performance is good. The packet loss rate of the system in this paper is significantly lower than that in reference [10] and reference [11]. In this paper, the system runs to detect the tampering areas of different types of telemedicine monitoring video images, and the packet loss rate is less than 1%; when the system in this paper runs to detect the tampering areas of different types of telemedicine monitoring video images, and the transmission delay is less than 200s. Figure 7 and Figure 8 verify that detecting the tampering area of telemedicine monitoring video image has high communication performance when the system is running.

Discussion
This paper studies the area tampering detection system of (2) Self-organization ability of network Because in the monitored area, the sensors are randomly deployed in the area, the location of sensor nodes can not be accurately set in advance, and the mutual neighbor relationship between nodes is not known in advance. For example, a large number of sensor nodes are sown into the vast virgin forest by aircraft, or placed in inaccessible or dangerous areas at will. In this case, sensor nodes are required to have the ability of self-organization, and can be automatically configured and managed, to automatically form a multi hop wireless network system for forwarding monitoring data through topology control mechanism and network protocol.
In the use of wireless sensor networks, some sensor nodes fail due to energy depletion or other factors. In this case, the topology of the network is required to change dynamically. The self-organization of wireless sensor networks should be able to realize the dynamic change of network topology.
(3) Wireless sensor networks are data centric. 13 EAI Endorsed Transactions on Pervasive Health and Technology 03 2022 -07 2022 | Volume 8 | Issue 31 | e3 Any application system based on wireless sensor networks is inseparable from the management and processing technology of sensing data. In sensor networks, each sensor node has the functions of both end node and router. The sensor node receives the query or control command of sink.
The core technologies are the technologies of data compression, data refining, data processing and data association based on sensor networks. Various implementation technologies of sensor networks must be integrated with these technologies.

Conclusion
In order to improve the security of remote medical video resources, a remote medical video region tamper detection system based on wireless sensor network is studied. The wireless sensor network is used as the communication technology of remote medical video region tamper detection to improve the application performance of remote medical video region tamper detection. Experiments show that the system can be applied to the actual remote medical video region tamper detection, and can effectively detect the tampered region in the remote medical video. It has high detection performance for deleting tampering, copying tampering and disordered tampering, and the packet loss rate of the system is significantly reduced. The packet loss rate of the system in this paper is less than 1%. Therefore, it fully shows that the proposed detection system based on wireless sensor network can better meet the requirements of remote medical video region tamper detection. However, the response performance of the system has not been verified in this study, so there is a problem of insufficient response performance. This problem will be studied in detail in subsequent studies.