Research And Implementation Of Auto-focusing Technology For High Definition Zoom Camera

The zoom camera in the field of security video surveillance has a history of nearly two decades.In recent years,with the advancement of digital image processing technology and chip technology,zoom cameras are being developed from analog to digital and standarddefinition to high-definition,and high-definition integrated cameras have emerged.With the advantages of high definition,flexible transmission mode,wide adaptation,compact size and convenient installation,it has quickly become the main force in surveillance cameras.As the key technology of the high-definition zoom camera,auto-focusing technology directly determines the image quality and image definition of the camera.The auto-focusing technology based on image processing is generally used in the HD zoom camera.It mainly includes three core sub-algorithms: the image sharpness evaluation function,zoom tracking algorithm and auto-focusing search algorithm.Based on the above considerations,this paper analyzes the existing problems of the auto focus algorithm,and proposes an improved algorithm from the perspective of engineering application.Finally,the comparison experiment of the algorithm is completed on the auto-focusing system.(1)Based on the analysis of the existing typical image sharpness evaluation function and Hisi-AF module,a comparative experiment was designed to demonstrate the feasibility of the selected Hisi-AF module in engineering practice.(2)This paper studies the current zoom tracking algorithm,aiming at the problems of narrow range of application and discontinuous zoom processing of the existing algorithms,the use of the state information of two adjacent zoom points to correct the evaluation tracking curve is proposed for the first time,so that the zoom tracking algorithm of this paper is obtained.Through comparative experiments,the zoom tracking algorithm proposed in this paper solves the problem of discontinuity in the zooming process of the feedback method and the adaptive method.At the same time,the zooming precision and zooming speed have also been significantly improved,the overall performance has been improved obviously,and the practical application value is high.(3)In this paper,different types of improved auto-focusing search algorithms have been studied.It is concluded that the existing algorithms cannot solve the focusing problem of light source scenes and single scene.Therefore,the auto-focusing search algorithm based on scene prediction is proposed,which can intelligently identify the scene and make a judgement.In the light source scene,the statistical value of the high brightness pixels of the image is used as the basis for the auto-focusing search.In the single scene,the author proposes an improved global search strategy.In the normal scene,when the focus motor is in the defocus area,the existing algorithm can not determine the direction of the peak of the focusing curve.The method of determining the peak direction by using the fuzzy degree evaluation model is proposed.The experimental results show that the algorithm designed in this paper improves the speed,stability and accuracy of auto-focusing search algorithm.(4)In the last chapter,this paper establishes auto-focusing system with processor Hi3518 A SoC chip as the core.The software and hardware design scheme of the system is described in detail.On this basis,the algorithm program and the motor driver are written,the experimental scheme is designed,and the experimental verification of the algorithm is completed.As a typical solution of high-definition zoom camera,the system has a strong market competitiveness.