Algorithm Improvement And Hardware Implementation Of Image Process Technology In Image Sensor

The image sensor chips are gradually widely used in many fields, like communication, security, medical treatment, and entertainment. As the core visual component, image sensor has an expansive market demand. Image sensor has special requirement in image processing, which aims to get the optimal visual effect. The noise in the image which generated by image sensor, will affect the quality of image negatively. It needs some effective noise reduction algorithm to optimize the image contaminated by noise.The traditional filtering algorithms don?ˉt have satisfied performance for hybrid noise, and it?ˉs difficult to implement some recently proposed algorithms with good image processing effect in hardware. In this thesis, at first a hybrid noise model is described. Then a new method of distinguishing noise and image edge is proposed. According to this method, a pattern-based adaptive hybrid filter is designed, which is easy for hardware implementation. And CIE94 color difference formulae, which can more veritably reflect human visual sense, is adopted as the criterion of image processing effect comparison. The simulation result shows the proposed algorithm has very good performance in the hybrid noise reduction with Gaussian noise and pepper and salt noise. And the hardware implementation and verification are also presented.The result of software simulation shows the effectivity of proposed algorithm. And the result of hardware implementation indicates the proposed algorithm can be easily integrated in image processing chip of image sensor, and meet the real time requirement.