Hybrid Error Control and Security Framework for Reliable Cornea and Infrared Video Communication in IoT Applications

In this paper, we present hybrid watermarking and error control techniques for reliable cornea and infrared frame communication through wireless networks in Internet of Things (IoT) applications. In the proposed watermarking technique, two stages of Singular Value Decomposition (SVD) watermarking are used. In the embedding stage, two watermark images are embedded in either the cornea or infrared frames using Block-based SVD (B-SVD) and SVD schemes, respectively. Then, the resulting watermarked cornea or infrared frames are transmitted through an erroneous wireless channel. At the receiver, the received corrupted cornea or infrared frames are recovered using a proposed hybrid post-processing error control technique. This technique comprises a Circular Spatial-scan Order Interpolation Algorithm (CSOIA) and a temporal Partitioning Motion Compensation Algorithm (PMCA) to reconstruct the erroneous cornea or infrared frames. Then, the Bayesian Kalman Filter (BKF) is utilized in the amelioration process due to its efficiency to smooth the remanent inherent corruptions in the formerly reconstructed frames to obtain high cornea or infrared video quality. After that, the watermark extraction stage is implemented to extract the watermark images from the watermarked frames. Simulation results on several cornea and infrared frames show that the suggested hybrid watermarking, and error control techniques reveal adequate subjective and objective video quality compared to those obtained with the traditional methods. In addition, the watermark detectability, robustness, and security are enhanced. Moreover, the experimental results show that the proposed watermarking technique is superior and more secure than the other previous techniques for embedding and extraction of watermarks, efficiently, in the presence of attacks.


Introduction
Today, the most important way to connect people all over the world is the Internet. It allows users to communicate and share multimedia data such as text, image, audio, and video, with low cost and high quality. In addition, the emergence of IoT has led to the need to send multimedia content over IoT networks. So, recently, the research efforts in wireless multimedia communication have increased and received more attentiveness. The multimedia communication over wireless networks or Internet has increased, dramatically [1][2][3][4][5][6]. However, one of the most serious problems that arise during multimedia communication is how to protect the critical data during the transmission process from illegal use [7,8]. Another serious problem of multimedia transmission over wireless networks or Internet is the induced errors on the communication channels [9][10][11][12][13][14]. Thus, to transmit the multimedia data over erroneous channels, efficient encoding must be implemented to maintain a good reception performance. Therefore, the video encoding prediction structure must depend on the time-and spatial-domain correlations inside frames to improve the decoding performance. Unfortunately, the highly-encoded videos are very sensitive to transmission channel errors.
Protecting confidential multimedia information is a moral and legal requirement. Data hiding is one of the protection ways in which part of the information is hidden into a multimedia object to authenticate it. This process is performed in such a way that the embedded data is not perceptible to an observer, but it is easily detected by the detector. Data hiding has two main categories: digital watermarking and steganography [15,16]. The insertion of a part of information into the multimedia content is called digital watermarking. This information should not be to a human eye, but can be detected with an appropriate algorithm. Figure 1 presents the general concept of the watermarking algorithm, which consists of the embedding and detection processes. This algorithm is suitable for several applications such as fingerprinting, copyright protection, and content authentication, because it has a very important feature that the content is inseparable from the watermark. The required characteristics of digital multimedia watermarking include imperceptibility, high robustness, high capacity, and large degree of security [17][18][19]. The general requirements for designing an efficient multimedia watermarking algorithm can be outlined as shown in Fig. 2. They are defined as follows: 1. Imperceptibility. The watermarked signal should be invisible for the human eye. In other words, the original host signal or image should not be affected by the embedded data.
2. High robustness. The watermarked signal should resist modifications such as removal or elimination of embedded data by attacks from unauthorized persons.
3. High capacity. The amount of embedded data in the host signal or image should be large as much as possible.
4. Large degree of security. The authorized person is the only one, who can detect the watermark. The multimedia content such as medical corneal and infrared videos to be streamed over wireless channels is subject to random and bursty packet corruptions due to packet dropping or fading-motivated bit errors [20][21][22][23]. So, errors might propagate to the neighbouring and the subsequent frames, resulting in inefficient video quality. It is intractable to retransmit the corrupted or erroneous frames due to delay restrictions on real-time video applications. So, there is a need for error control techniques such as post-processing concealment techniques at the receiver side. The concealment techniques are recommended, as they decrease the channel errors without increasing the delay or requiring any complicated encoder adjustments [4,9,11]. The cornea is the transparent layer that faces the light, which enters the eye. Cornea and Sclera form the outer tunic of the eye and are mechanically strong. They protect the eye from any foreign object. The cornea has five main layers, namely epithelium, Bowman layer, Stroma, Descemet membrane, and endothelium. The main function of the endothelium is to pump the excess water out of the stroma to preserve its mechanical structure and optical clarity. The corneal endothelium has an important impact on the human vision because of the direct relationship between the healthy cornea and the endothelium cells [24][25][26]. In this paper, we will transmit and evaluate corneal frames over wireless networks for telemedicine applications. In addition, we will investigate also the transmission of infrared video frames over wireless networks for further analysis. Both types of video frames are considered in IoT applications.
Several error control techniques are suggested to minimize the effect of packet damages during the video transmission process.
In the state-of-the-art works [27][28][29], the Automatic Repeat Request (ARQ) and Forward Error Correction (FEC) algorithms have been proposed. Unfortunately, they increase the transmission bit rate, and they are not recommended for real-time video applications. Therefore, in this paper, we propose efficient post-processing error control techniques at the decoder, which neither add redundant data nor increase the transmission bit rate. The proposed post-processing error control techniques in this work have the feature of enhancing the received video quality without any modifications in the encoder software or hardware or in the transmission bit rate.
These techniques depend on the time and spatial correlations inside the corneal and infrared sequences to conceal the corrupted frames. At the decoder, a hybrid approach of CSOIA and PMCA is discussed in section 3.2 in order to determine the Disparity Vectors (DVs) and Motion Vectors (MVs) of the corrupted video frames. After that, for more enhancement of the concealed corneal and infrared video frames, the BKF is applied on the recovered DVs and MVs to enhance the overall video quality.
The main contributions of this paper are summarized as follows: 1. A hybrid framework comprising error control coding and watermarking is introduced for reliable transmission of corneal and infrared video frames over a wireless communication channel.
2. The proposed hybrid framework achieves superior performance compared to the traditional algorithms in mitigating the effects of communication channel errors and multimedia attacks.
3. An efficient hybrid algorithm based on block-based SVD watermarking and traditional SVD watermarking is introduced. This hybrid algorithm improves the copyright protection against severe attacks.
4. The proposed hybrid watermarking algorithm improves the capacity by embedding two different watermarks without affecting the imperceptibility and visual quality of the watermarked frames. This algorithm achieves an acceptable performance of watermark detection compared to the traditional algorithms.
5. The proposed hybrid error control techniques have a great effect on improving the overall performance of the proposed framework by achieving superior performance in watermark extraction.
The rest of this paper is organized as follows. Section 2 introduces the related work and the basics of the SVD watermarking process. In section 3, we present the proposed hybrid watermarking algorithms, the proposed error control techniques for corneal and infrared frames, and the suggested BKF amelioration technique. Simulation outcomes and comparative discussions are summarized in section 4. Section 5 presents the concluding remarks.

Traditional SVD Watermarking
From the image processing perspective, the SVD is a mathematical technique which ensures good stability for Singular Values (SVs). When small changes occur in the watermarked image, its SVs do not change majorly. Hence, SVD watermarking is robust against different attacks. The algebraic properties of an image can be represented by the SVs of that image. Using these properties of the SVD of an image, the watermark can be embedded in the SV matrix without large variations in the watermarked image. Liu et al [30] introduced a watermarking algorithm based on the spatial-domain SVD. Recently, watermarking schemes based on SVD have gained popularity due to the very simple implementation process and the attractive mathematical features of the SVD.
In the SVD, any matrix B may be written as [30]: where X and Y are orthogonal matrices such that T  X X I , T  Y Y I , and I is an identity matrix. S is a diagonal matrix containing the SVs of the matrix B . The columns of X are the left SVs of B , and the columns of Y are the right SVs of B . Figure 3 illustrates the watermark embedding process for the traditional method of Liu [30].

Embedding Process:
The steps of watermark embedding are as follows: 2. The matrix D is obtained by adding the watermark ( W matrix) to the SVs of the original matrix.
3. The new modified matrix ( D matrix) is decomposed by SVD.
4. The watermarked image is obtained as follows: Fig. 3. The SVD watermark embedding process.

Detection/Extraction Process:
Having w X ,

S,
w Y matrices and the possibly distorted image, the corrupted watermark is extracted from the watermarked image as follows: 2. The matrix D * is obtained.
3. Using the D * and S matrices, the watermark is obtained.
The corruption from attacks is referred to with * .

Proposed Hybrid Security and Error Control Framework
In this section, the proposed hybrid watermarking and error control framework is introduced. The general framework of the proposed cornea and infrared video communication system is shown in Fig. 4. Firstly, the cornea and infrared frames are watermarked using the proposed embedding process to embed two watermarks in the host cornea and infrared frames. After that, the watermarked cornea and infrared frames are transmitted through a heavily corrupted wireless channel and received at the receiver side. Due to the channel noise effect on the transmitted watermarked cornea and infrared frames, they are affected by the wireless channel errors. So, we propose efficient post-processing error control algorithms at the receiver to conceal the errors in the received watermarked cornea and infrared frames, efficiently. The proposed error control algorithms have the advantage of neither needing modifications at the transmitter side nor increasing the transmission bit rate. After performing the error recovery process of the received corrupted watermarked cornea and infrared frames, the proposed extraction process is implemented on the resulting concealed watermarked cornea and infrared frames to extract the watermarks. The suggested embedding and extraction processes, the proposed post-processing error control schemes at the receiver side for the corrupted watermarked frames, and the proposed BKF enhancement algorithm will be discussed and explained in detail in the next sections.  Fig. 4. General framework of the proposed hybrid security and error control framework for wireless cornea and infrared video communication.

Proposed Hybrid Watermarking Algorithm (B-SVD/SVD)
The proposed hybrid watermarking algorithm is based on the SVD and the BSVD. It satisfies the following watermarking requirements: 1. Improving the level of security of the algorithm.
2. Maintaining the imperceptibility, where the quality of the original image is not affected when increasing the capacity of information embedding.
3. Enhancing the watermark detectability. The basic idea of the proposed algorithm is embedding two watermarks in the original cornea or infrared frames by using a hybrid B-SVD and SVD structure to produce the watermarked cornea and infrared frames. Briefly, the proposed hybrid watermarking algorithm consists of embedding and extraction processes as shown in the general block diagrams of Figs. 5 and 6.
As shown in Fig. 5, to get the final watermarked frames, the embedding process consists of two stages. In the first stage of the embedding process, watermark 1 is embedded into the original cornea or infrared frame by using the B-SVD watermarking algorithm. Then, watermark 2 is embedded into the primarily watermarked frame by using the SVD watermarking algorithm. Figure   6 shows the extraction process, which consists of two stages. Firstly, watermark 2 is detected and extracted from the final watermarked frame. After that, watermark 1 is extracted in the second stage.

Proposed Embedding Process:
In the proposed embedding process, we apply the B-SVD watermarking and the SVD watermarking algorithms, respectively, as shown in Fig. 8 with the following steps: 1. The original cornea or infrared frame ( A matrix) is divided into non-overlapping blocks with the same size, and then the SVD is applied on each block (Ci matrix) to obtain the SVs ( i S matrix) of each block, where 1, 2, 3. . . , i N  and N is the number of blocks.
2. The watermark image ( W matrix) is added to the S matrix of each block.
3. The SVs of each wi S matrix are obtained by applying the SVD on each Di matrix.
4. The watermarked blocks in the spatial domain are built using the SVs of each i 5. The watermarked frame in the time domain ( wp A matrix) is built by combining the watermarked blocks back into one matrix.
From the previous steps of the first stage of the proposed algorithm, we get the primarily watermarked frame which is the input to the second stage having the following steps: 6. The SVD is applied on the primarily watermarked frame.
To obtain the matrix D, the watermark ( W matrix) is added to the SVs of the primarily watermarked matrix, and then the SVD is applied.
T w wp  D X S Y (16) 8. Using the modified matrix ( wp S matrix), the final watermarked frame ( Figure 8 shows the extraction/detection process through the application of the SVD and B-SVD extraction algorithms, respectively. The proposed extraction process consists of the following steps:

Proposed Extraction Process:
1. The SVD is applied on the received final watermarked frame ( 2. The matrix containing the watermark is estimated.
3. The matrices  D and wf S are utilized to extract watermark 1, which may be corrupted.
To get the watermark 2, by knowing

Proposed Spatial and Temporal Post-processing Error Control Algorithms
The temporal Outer Block Boundary Matching Algorithm (OBBMA) and the spatial Weight Pixel Averaging Interpolation (WPAI) techniques [31][32][33][34][35] were proposed for 2D video concealment. However, they fail and cannot work efficiently in the scenario of severely corrupted channels, and therefore they result in deterioration of the communication performance. Thus, we present an enhanced approach of temporal PMCA and spatial CSOIA procedures that proved efficiency in previous works [4] for efficient three-dimensional video transmission. So, in this work, we exploit both of them in the proposed framework for concealing the cornea and infrared frames for achieving an adequate error control performance. see [4]. Therefore, we use CSOIA and PMCA techniques in the current proposed work due to their advantages in mitigating the channel errors.

Proposed BKF Enhancement Algorithm
The proposed BKF enhancement algorithm is introduced in this section. It proved its efficiency in the previous work [9,11] for efficient three-dimensional video transmission. In this work, it is exploited for smoothing the optimum formerly-determined DVs and MVs by the proposed spatial and temporal post-processing error control algorithms. It is used for compensating for the remnant Side information S measurement corruption among the chosen candidate DVs and MVs. Thus, the BKF is used as an amelioration stage for enhancement of the cornea and infrared frame visual quality. Therefore, the stages of the proposed BKF algorithm [9] are iterated until the whole erroneous frames are recovered.
Consequently, if the initially-determined DVs and MVs are not sufficient for an optimum error control process because of limited available surrounding pixel information and severe errors. Therefore, they are then enhanced by the predicted DVs and MVs, which are estimated through the utilization of the suggested BKF technique to achieve better cornea and infrared video transmission performance. More theoretical descriptions and mathematical details about the BKF algorithm can be found in our previous work in [9,11]. Therefore, we exploit it in the current proposed work due to its benefits and impact in lessening the communication errors.

Simulation Results
In order to appraise the efficiency of the suggested hybrid algorithms, we have carried out different simulation tests on cornea and infrared frames with and without attacks. For each video, the encoded cornea and infrared frames are obtained, and after that, they are watermarked using the proposed embedding process. After that, they are transmitted over a lossy wireless transmission channel, which is simulated using the Gilbert channel model [34]. Then, the received watermarked and corrupted cornea and infrared frames are subsequently decoded and concealed by the suggested post-processing error control techniques at the receiver side. After that, the proposed extraction process is employed on the concealed frames to extract the possibly corrupted watermarks. To assess the sufficiency of the suggested error control schemes, the PSNR, the Feature Similarity (FSIM) index, and the Structural Similarity (SSIM) index [35] are used as objective metrics to determine the quality of the reconstructed and concealed cornea and infrared frames. Further details about their definitions and equations can be found in [36]. Collect   To assess the efficiency of the proposed embedding/extraction processes, the visual results ensure watermark invisibility and no degradation in the watermarked frame quality compared to the original frames. Also, the PSNRs of the watermarked frames and the correlation coefficients (Cr) of the extracted possibly corrupted watermarks with the original ones are estimated. The large PSNR value is considered as an imperceptibility quality metric. It should be infinite, but this is not possible for watermarked frames. The mathematical representation of the PSNR is given in (24), where () x, y A is the original frame and   w x, y A is the watermarked frame. The Cr value is considered as a robustness quality metric that measures the difference between the extracted and the original watermarks. The closer the Cr value to one, the larger the robustness is [37,38]. The mathematical representation of the Cr is given in (25) Fig. 10. The cornea and infrared tested host frames.
In our simulation experiments, we apply different types of multimedia attacks on the watermarked frames, and then extract the watermark to test the robustness of the proposed watermarking algorithm. We have run two experiments for the tested cornea and infrared videos. The first experiment on the cornea video was implemented on frames 15, 20, and 25. The other experiment on the infrared video was implemented on frames 10, 30, and 50, as shown in Fig. 10. Figure 11 presents the utilized watermark images.  To clarify the efficiency of the proposed hybrid watermarking algorithm in protecting and securing the transmitted cornea and infrared frames in the presence of attacks, we have compared its performance, robustness, and imperceptibility with those of the state-of-the-art watermarking schemes such as SVD and DWT + SVD [37][38][39][40]. In the introduced simulation results, the B-SVD + SVD refers to the proposed hybrid watermarking algorithm.
In Tables 1, 2 [37][38][39][40] in the presence of different attacks that the proposed hybrid watermarking algorithm gives not only the highest PSNR values between the original and watermarked frames, but also the best correlation between the original and extracted watermarks. Therefore, it is clear from the presented results that there is a higher similarity between the original and watermarked frames. The proposed algorithm also improves the similarity between the original and extracted watermarks. Moreover, the proposed algorithm achieves high PSNR and Cr values for all tested cornea and infrared frames.   Table 5 shows the average PSNR, FSIM, and SSIM of the proposed algorithms compared to the algorithms in [12][13][14][32][33][34][35] at PLR=40%. It can be seen that the proposed algorithms outperform the state-of-the-art ones. So, it is observed that with the proposed error control algorithms, the recovery quality can be further enhanced by exploiting the hybrid structure of CSOIA, PMCA, and BKF algorithms. Therefore, the significance of exploiting the proposed error control algorithms for reconstructing the corrupted cornea and infrared frames sufficiently is clear in the scenario of severelycorrupting channels. From all presented simulation results on the tested cornea and infrared frames, we perceive that the suggested post-processing error control algorithms always achieve superior PSNR, FSIM, and SSIM values. Also, it is important to reveal the effect of the utilization of the error control algorithms in the watermark detection process. To demonstrate the impact of employing the suggested error control algorithms in efficiently extracting the watermarks from the watermarked infrared and cornea frames, we perform different simulation tests to check the performance of watermark extraction with and without using the proposed hybrid error control algorithms for the cornea and infrared frames. Table 6 presents the average Cr simulation results of the extracted watermarks from the watermarked cornea and infrared frames with and without exploiting the proposed hybrid error control algorithms in the presence of different types of attacks with a packet loss rate of 40%. The significance of exploiting the proposed error control algorithms for a higher possibility of watermark extraction with higher correlation coefficient values from the corrupted cornea and infrared frames is clear in the scenario of attacks and severely-corrupting channels as in the presented state of PLR=40%.

Conclusions
This paper presented an efficient hybrid error control and watermarking algorithm for reliable cornea and infrared frame communication through wireless channels. The evaluation metrics for the proposed watermarking algorithm includes stability, reliability, and robustness. Experimental results revealed the superiority of the proposed hybrid watermarking algorithm in maintaining high robustness and fidelity in the presence of multimedia attacks compared to the existing watermarking algorithms.
Moreover, the simulation results proved the importance of the suggested post-processing error control techniques on recovering the erroneous cornea and infrared frames for achieving significant PSNR, SSIM, and FSIM results. From all obtained results, we also ensured the importance of employing both proposed error control and security techniques for efficient transmission of medical data and infrared frames over wireless channels for secure and reliable telecommunication and IoT applications.