Research On Image Stitching Technology Of Compound Eye Lens

With the development of technology and society,people's demand for large field of view and high resolution images has become increasingly prominent.Compared with the traditional single eye optical imaging system,the bionic compound eye vision system constructed by imitating the compound eyes of insects has unique advantages such as smaller size,larger field of view,and more sensitive response to moving objects.The use of image stitching technology to synthesize sub-image sequences with overlapping regions collected by the bionic compound eye vision system into a large field of view image has become a hot research topic in machine vision.This paper designs and builds a bionic compound eye vision system,and proposes an improved image stitching algorithm based on the characteristics of compound-eye imaging to complete the synthesis of images with a large field of view.The research work of this paper is mainly summarized as the following aspects:1)Design and construction of bionic compound eye visual system.Due to the large size and high cost of the compound eye vision system constructed by the camera array,which is not conducive to integration.This paper proposes a method for building a bionic compound eye vision system using a compound eye lens.The system is mainly composed of a compound eye lens,a relay system,and an image sensor.The relay system transmits the light information collected by the compound eye lens to the image sensor.By adjusting the distance between the compound eye lens and the relay system or the focal length of the relay system,the clear imaging on the image sensor can be guaranteed.2)Calibration of the bionic compound eye vision system.The problem of image distortion in the imaging process of compound eye vision system is unavoidable,so it is necessary to calibrate the bionic compound eye visual system for image correction.Different from the calibration of monocular vision system,the calibration of compound eye vision system requires not only the calibration of each sub-eye but also the global calibration of the position relationship between sub-eyes.In this paper,Zhang Zhengyou's calibration method is used to calibrate each sub-eye of the bionic compound eye visual system,and the global calibration of the compound eye vision system is completed by using the public view of several sub-eyes and using the plane target as the intermediary.3)Panoramic image stitching of bionic compound eye vision system.This paper,on the basis of detailed analysis of the relevant theories of the bionic compound eye visual system image stitching,analyzes the registration performance of the commonly used three image registration algorithms,which concludes the SIFT algorithm has best performance but lowest efficiency.Further,the SIFT algorithm is improved from two aspects.On one hand,the overlap area between images is used to reduce the search space,and on the other hand,the BRIEF descriptor is replaced to improve the matching efficiency.The experimental results show that the improved algorithm not only has a good stitching effect and robustness,but also can speed up the stitching speed by more than 30%.Finally,the improved algorithm is used to complete the large field image synthesis of the compound eye sub-image sequence.The resolution of single sub-eye was 380*380 before stitching and 1038*1038 after stitching.The SIFT algorithm spends 25.23 s,while the improved algorithm spends 15.39 s,and the stitching efficiency is increased by 39%,which can achieve the expected large field of view.In summary,this paper designs and constructs a bionic compound eye visual system,completes the calibration of the vision system,and proposes an improved stitching algorithm based on the imaging characteristics of bionic compound eye.The algorithm has improved the real-time performance of stitching on the basis of ensuring the quality of subimage sequences stitching.