Dehazing algorithm for complex environment video images considering visual communication effects

Abstract

The communication effect of outdoor computer vision systems is closely related to the dehazing performance of video images. Currently, video image restoration methods in complex environments are still in an urgent development stage. By analyzing the characteristics of smoke, considering the local and global consistency of clean videos, as well as the sparsity of smoke videos, and elaborating on low rank tensor recovery, a smoke removal model is constructed in complex environments. The alternating direction multiplier optimization algorithm is used to solve the dehazing model. The dehazing algorithm for video images in complex environments had ideal dehazing effects in different types of images. However, the dehazing effect of other dehazing algorithms was not particularly significant. The maximum iteration times for different video image dehazing algorithms were 30, 40, 65, and 80, respectively. The maximum resolution of the algorithm also showed a gradually decreasing trend with the increase of SNR. For the dehazing algorithm of video images in complex environments, the optimal region scales corresponding to the three target scales were 3.2, 3.3, and 4.2, and the accuracy obtained was 90.5%, 90.6%, and 90.8%, respectively. The normal range of SSIM for video image dehazing algorithm in complex environments was [-1,1]. The normal value of PSNR was 30–40 dB. The PSNR and SSIM of the four types of video images meet the requirements. The constructed complex environment video image dehazing algorithm provides valuable suggestions for the development and improvement of computer vision systems. The constructed complex environment video image dehazing algorithm greatly improves the dehazing performance of video images, thereby greatly enhancing the communication effect of outdoor computer vision systems.