Research On Auto-focusing Method Of Photoelectric Measurement Equipment Based On Image Processing

Sharpness image is the important guarantee for photoelectric measurement equipment to record the trajectory data, flight attitude and actual scene. However, the relative distance between photoelectric measurement equipment and target is almost changing, leading to defocus of target image surfaces. Target imaging blur caused by defocus is an inevitable problem in photoelectric measurement, hindering the extraction and analysis of image information, decreasing the measurement accuracy of trajectory data and flight attitude. Auto-focusing and imaging sharpness is essential for improving the performance of photoelectric measurement equipment. The traditional focusing method of photoelectric measurement equipment usually achieve manual focusing according to the observation of human or measures the distance of target to achieve focusing according to the distance. It requires a higher hardware platform and costs larger, but achieves low precision of focusing. Therefore, the auto-focusing problem of photoelectric measurement equipment is urgent to be resolved.From the perspective of image processing, this paper deeply researches on the auto-focusing method based on image process, through focus window construction, image quality assessment and searching for focal position. Combining with the practical engineering requirements, this paper proposes an auto-focusing algorithm, which is suitable for the photoelectric measurement equipment.Focus window construction is the first step in auto-focusing process, refrain the affect of background and d reduce the amount of calculation. At the same time, it determines the target area for the following steps. Since the position and size of target changes from time to time, we proposed an adaptively Focus window construction method from theperspective of target tracking. The method obtains the position of target using Mean-shift algorithm, and conduct the visual saliency detection to determine the target area. Then, the adaptive focus window is built by spreading the boundary of target area, and adaptive kernel function is designed for Mean-shift algorithm combined with Epanechnikov function, reducing the influence of background and the changes of target scale, achieving the favorable target tracking.Image quality assessment(IQA) is a critical step for auto-focusing, and the IQA result determine the precision of focusing. According to the changes of target, this paper proposes a novel no reference IQA method based on non-subsample shearlet transform(NSST) and natural scene statistics(NSS). It extracts the NSS features in the shearlet domain, and achieves the image quality by using support vector regression. The method can overcome the effect of content and has a strong correlation with human perception. Meanwhile, it can judge the defocus extent just by only one frame, and is suitable for rough focusing. In order to ensure the precision of focusing, this paper also proposed a novel no reference IQA method for images with slight blurred using double Gaussian reblur and singular value decomposition(SVD). It measures the changes of similarity and energy to achieve the image quality. The IQA result is susceptive to the quality of images with slight blurred, and is suitable for fine focusing.Searching for focal position is the final step for focusing. This paper improves the hill climbing search method and proposes a searching method for the auto-focusing of photoelectric measurement equipment. In rough focusing, the method of NSST and NSS is applied for IQA, and the searching step length is optimized using improved hill climbing search method. In fine focusing, the method of double Gaussian reblur SVD is applied for IQA, and the hill climbing search combined with curve fitting is used to search focal position.Based the proposed auto-focusing method, we design a hardware platform used for achieve auto-focusing of photoelectric measurement equipment. After a plenty of focusing experiments, the paper analyses analyze the focusing accuracy, stability, real time feature and the source of error in detail according to the focusing results. The results show that the program is feasible and useful for engineering. The experiments results show that the auto-focusing system achieves the auto-focusing of photoelectric measurement equipment effectively, and provides a focus solution for practical application.