Research And Realization For Bionic Eye Movement Control

The human visual system is an important sensory to capture external information which has an extremely intricate physiological mechanism in target tracking, image acquisition, understanding, etc. Machine vision can make the machine sense the environment by means of visual information, enhancing the machine’s ability to adapt to the environment as well as improving its intelligence. Characteristics and mechanism of the human visual system can give a lot of inspiration to research on machine vision and applying these ideas to the engineering field can improve the visual perception ability of the machine. So it is a meaningful work. The bionic eye research in this paper focuses on bionic eye movement control and the main work is as follows:This paper first analyzes the bionic eye research background and significance, and summarizes the current research work and point out the main problems of them, and introduces the main content and structure of this article.Secondly, The human eye movement form and characteristics are analyzed and on this basis, a hardware system platform and a software simulation platform are set up for demonstrating and validating algorithms and models in this paper. Bionic eye hardware platform can complete image acquisition and processing and simulation of eye movement and head movement. The software simulation platform coded by Matlab can demonstrate the movement of head and eye in the process of target tracking vividly with animation.Thirdly, an improved RLS algorithm and ILC are adopted to implement the smooth pursuit. An improved RLS algorithm is used for prediction of velocity for sinusoidal motion, uniform motion, uniformly accelerated motion, and simple pendulum movement with large angle taking air resistance into account and the image processing delay can be offset. The improved RLS algorithm can adapt to different forms of movement quickly as well as resist noise effectively. ILC is adopted to realize the smooth pursuit of the periodic movement and it can achieve accurate tracking performance after several cycles of learning. The tracking effects of two algorithms are verified by experiment.Fourthly, the saccade control algorithm is studied. Human physiological model of saccade is analyzed and on this basis, the saccade control model is researched. This paper deduces the formula of calculating the gaze shift amplitude according to the image coordinate of target based on the camera imaging model. A red ball can be recognized in RGB color space and fast algorithm is used for color space conversion, ensuring the recognition accuracy and real-time performance. The saccade model is verified by simulation and the bionic eye hardware system.Finally, we summarize the work of this paper and propose the future research direction.