Research On Adaptive Error Concealment Algorithm Of Stereoscopic Images And Videos

With the development of the Internet,the application scene of stereoscopic videos becomes more and more abundant.However,due to its large amount of data,methods of data compression and coding,stereoscopic videos poses a huge challenge to the transmission network.Once a bit error or packet loss occurs in a network transmission due to various reasons,it will have a huge impact on the quality of the videos received by the user.Therefore,it is very important to control the error of stereoscopic videos transmission.Error concealment is one of the main methods of the error control of stereoscopic videos.This paper aims at the error control problem of stereoscopic videos,analyzes the characteristics of stereoscopic images and stereoscopic videos firstly,and then proposes error concealment algorithms for stereoscopic images and stereoscopic videos respectively.The specific research contents are as follows:According to the correlation between viewpoints of stereo images,this paper proposes an error concealment method based on superpixel segmentation and adaptive disparity selection strategy.First,the superpixel segmentation label of the left image and right image are acquired.Then the superpixel segmentation label are added to the constraint conditions in the disparity search matching process,so that the spatial organization correlation of the pixels in the image is included in the disparity estimation.In order to reduce the randomness of disparity selection in certain regions of the disparity selection strategy,this paper also proposes an adaptive disparity selection strategy to improve the optimal disparity selection.After setting the specific gravity threshold of candidate disparity,all disparities with a cumulative proportion less than the threshold are added to the candidate disparity set.And then the best estimated disparity is determined by the boundary smoothness principle.Finally,recovering the lost image.The experimental results show that this algorithm has certain advantages in both objective and subjective quality evaluation.Stereoscopic videos are closely related to stereoscopic images.And they have correlation spatial correlation,inter-view correlation,and inter-frame correlation.Aiming at these characteristics of stereoscopic videos,this paper proposes a hybrid spatial and temporal error concealment method for stereoscopic videos based on image pyramid theory.First,downsampling is performed on two adjacent frames of the viewpoint of the non-lost block of the stereoscopic videos,and then the scene-switching detection is performed using the frame difference method on the down-sampled videos frames.For the scene switching frame,the stereoscopic image error concealment algorithm is used to process the videos of the lost viewpoint.For non-scene switching frames,the motion characteristics of lost blocks in the video frames with missing blocks are further detected.The motion characteristics of the lost block are comprehensively evaluated using the image blocks around the lost block and the image blocks in the corresponding area in the downsampled image of the videos frame.The motion characteristics of lost blocks are divided into static,relative motion and strenuous motion.The missing blocks of different motion characteristics use the corresponding method respectively for error concealment.Experimental results show that compared with other algorithms,the proposed algorithm has certain advantages in the evaluation of objective quality.At the same time,the proposed method preserves the details and texture of the image better in the evaluation of subjective quality,and it has low complexity.