Research On The Auto-focus Algorithms Based On Image Processing

Auto focusing technology is an important technology of imaging system.In the imaging system,it is a priority to achieve accurate focusing,which is related to the imaging quality of the system.In this thesis,an optical imaging system model has been estabished,which is helpful to analyze the cause of blurred image in the defocus case,and simulate the point spread function of the imaging system with Gauss model,which lays a theoretical foundation for the research of auto focusing technology.Compared with the traditional auto focusing technology,image information can be obtained only according to the auto focusing process for automatic focusing technique based on image processing without additional optical equipment,and better research value and development prospects are reflected.The main contents are summarized as follows:An improved Sobel algorithm of image definition evaluation based on segmentation algorithm has been proposed,considering that the airspace class definition evaluation function is easily affected by noise shortcomings.The main part of the method is taking the gradient function,making full use of the characteristics of spatial function is simple and easy to realize the real-time focusing;through the two diagonal template increase the direction of the Sobel operator,it improves the ability of operator to extract the gradient information in different directions;image from background part is susceptible to noise,the robustness of the algorithm is enhanced by using the threshold segmentation of the gradient image and discarding some images of background is susceptible to noise pollution.For climbing search method is easy to fall into local extremum and the shortcoming that searching the peak by move camera lens repeatedly,an improved hill-climbing method based on curve fitting method has been proposed.The search step by adaptively in the coarse focusing stage uses the long step not only improve the search efficiency,but also reduce the interference of local extremum the flat area evaluation curve;curve fitting value can rapid position the best focal position through the evaluation stage of the collected fine adjustment function,reducing the search time spent in searching the focal position.The feasibility of the improved algorithm is verified by our experiments,in which the improved definition evaluation algorithm is compared with the traditional algorithm.The performance index of the evaluation function is analyzed quantitatively and qualitatively.The experimental results show that the improved algorithm proposed in this thesis meet the needs of the camera auto focus system based on image processing,when the focusing effect is stable and the efficiency of focusing is improved.It has significant instruction meanings for engineering.