| The traditional video-aware system can not meet the application requirements in the wild environment where is no power and communication network infrastructures.While wireless video sensor network has characteristics of self-organization,low power and so on that can be a good solution to this problem.However,the field scene is usually shrouded in haze weather conditions,resulting in the collection of video images blurred,and it is difficult to get the details of the scene information.To this end,this paper designs and implements a wireless video sensing system based on defogging processing.Firstly,this paper carries on the demand analysis to the system based on the image defogging and the wireless sensor network technology according to the requirements of the video captured in the field environment.It designes the system overall structure and the node architecture,and introduces the hardware and software implementation of each function module of the system in detail.Secondly,this paper proposes an improved single image defogging algorithm for the defects of Tarel’s algorithm.The tatal variation model is used to replace the median filter of the original algorithm,which can preserve the edge characteristics of the image on the basis of greatly improving the computational efficiency.Aiming at the failure of the estimated atmospheric veil in the bright region,a fault tolerant mechanism based on haze density feature is proposed,which improves the universality of the algorithm.The experimental results show that the proposed algorithm can effectively preserve the edge characteristics of the restored image that is clear and natural.it also has certain advantages of defogging effect and efficiency over the current main defogging algorithms.Then,this paper proposes a fast video defogging algorithm based on the improved single image defogging algorithm.According to the association between the video frames,the video sequence is defogged in image groups.For each image group,the restoration of the key frame uses the single image defogging algorithm mentioned above,and the rest of the non-key frames are based on the motion vector.Therefore,it can greatly reduce the defogging redundancy and ensure the real-time performance of the video defogging.The experimental results show that the contrast and clarity of the restored video image are improved obviously,and the video stream can be outputin real time.Finally,this paper implements the prototype system combined with the video defogging algorithm mentioned above.The results of functional test and performance test show that the wireless video sensing system based on defogging processing can meet the requirements of field environment and it is portable. |
