Research On Electronic Image Stabilization System Based On Complex Motion Video

In recent years,with the progress of science and technology,video captured by camera is not only used for people's viewing,but also widely used in people's livelihood,military and other fields.More and more machines are equipped with camera devices.Video recorded on different devices will have unreasonable applications,such as video captured by camera in automatic driving to monitor road information,video camera at parking lot entrance and exit to detect car license plate number and so on.All these premises require that the device has a clear and stable input video,while video recording will produce jitter,irregular jitter.The interference algorithm is very important for the recognition of the information in the input video,so the image stabilization of the video is very important.In this paper,the key technologies of motion estimation and artificial bee colony algorithm used in electronic image stabilization are studied based on the background of complex motion video with random step size and rotation jitter.The main contents are as follows:Firstly,in order to solve the problem of slow convergence speed and easy to fall into local minima of traditional bee colony algorithm,the idea of differential evolution is used for reference.In the initial stage of optimization,by introducing multiple random values,the diversity of population is maintained and the premature algorithm is prevented.In the later stage of search,stochastic parameters are reduced,and the optimal value is added as the guiding parameter to improve the local search ability of the algorithm and accelerate the convergence speed of the algorithm.Degree.This method not only prevents the algorithm from falling into local optimum,but also improves the speed and accuracy of optimization.Secondly,aiming at the problem of long operation time and low efficiency of traditional algorithms for recognizing complex motion video,a two-block matching method is proposed.The whole frame image is scaled and moved by two motion vectors,and the rotational jitter recognition of video is completed.The feasibility of the algorithm is verified by experiments.Thirdly,aiming at the problem of low search efficiency caused by the irregular displacement step and motion direction of some frames in complex motion video,the improved bee colony algorithm is introduced as the search strategy of block matching method,and the motion vectors of the first two frames are used as the prior knowledge of the next operation to form an initial population based on prior knowledge,so as to achieve faster image stabilization of complex motion video.Finally,a complete electronic image stabilization system is established,and the improved algorithm is evaluated by comparing different types of jitter videos.The results show that the improved algorithm can perform fast and accurate image stabilization for complex motion videos.