Multi-channel Information Infusion Of Spherical Compound Eye

Compound eye is more and more concerned by academia and industry, due to its advantage of large view field, high agility, and muti-channels. Therefore scholars of many countries begin to focus on their interests into the research of artificial compound eye. However, the advantage of compound eye is usually limited because of the restriction of fabrication technique. The structure of planer compound eye limits its field of view. Although spherical compound eye is very similar with the nature compound eye in large field of view, the planer image sensor requires that every eyelet lens in the spherical surface has different focus length and different profile which is usually aspherical. Furthermore, the aspherical lens is very difficult to fabricate by using traditional fabrication technique.Our lab designed a new artificial compound eye model based on the nature compound-eye structure. This model handles the contradiction between the spherical dome and planer image sensor. Its optical performance can be optimized based on the liquid lens fabrication system developed by our lab. In this work reported, the optical performance of this model is analyzed. We propose the technique of image denoising and image transferring in the image preprocessing section, including sub-pixel method and Neural network arithmetic which solve the problem of aberration and noise in compound eye imaging process. Moreover, we propose multi-channel information infusing arithmetic, which contains image mosaic, super resolution imaging and moving object detection.To examine the feasibility of the arithmetic in image preprocessing and multi-channel information infusing, we performed simulation by using Matlab, and built the experimental platform for further observation. Both of results show that neural network has good performance in the nonlinear calibration and the center of gravity method can record the position of spot precisely. Besides, the results also show that although compound eye is an optical system that sacrifice its spatial resolution to improve its field of view, our compound eye model can still detect the precise position of moving object. Our model will have has good application foreground.