Camera Autofocus Algorithm And Its Application

In camera ISP(Image Signal Processing) key technology plays an important role in image processing. In this paper, we study the basic principle of camera imaging, and a series of image preprocessing procedure, which from the image acquisition base on the image sensor to video coding. Based on the above, this paper study in detail the automatic focusing algorithm and depth estimation algorithm, and an improved auto focusing algorithm and a depth estimation algorithm are proposed in this paper.AF(Auto Focusing) is the process of attach the best of the image on the camera focal point by adjusting the position of the focal point. In this paper, a new automatic focusing algorithm for high definition video conference is proposed. Through the study the video conference scene analysis, the auto focusing algorithm is divided into three modules, the auto focusing preparation module, the auto focusing module, the scene monitoring module, through loop control camera lens position in this three modules, the camera lens can be quickly and stably move to the focal point. In this paper, the main problems encountered in the process of designing adaptive auto focusing algorithm are described, and the corresponding solutions are given. The experimental results show that the auto focusing algorithm designed in this paper can be real-time, stable and fast auto focusing, and improve the quality of the image in camera.The defocus image depth estimation algorithm(Depth from Defocus, DFD)is according to the relationship between image plane circular fuzzy spot radius on the camera focal and the object distance in optical system to estimate the depth information in defocus image. Through Canny edge detection algorithm detected the edge information of the image. Calculated the ratio of the fuzzy value of the edge of the image and the value of the edge of the blur image, attach the depth information of the defocus image. In this paper, an improved depth estimation algorithm is designed to solve defocus image in depth estimation of coke confusion problem under camera large times. By the divide the front and rear area of the defocus image, according to the fuzzy mean value in each region to determine the focus area in defocus image, for different focus regions, the all pixel points of the image are optimized fuzzy degree value by different method, the image depth estimation is obtained. The experimental results show that the improved algorithm can effectively eliminate coke confusion problem of depth estimation in defocus image taken from camera large times, the depth estimation is not affected by the focus point move, and the depth information of the defocus image can be estimated correctly. This algorithm has good robustness.