An Adaptive-Fuzzy Filter Algorithm for Vision Preprocessing

There are Gauss noise and impulse noise in the great mass of the input images of the robot vision system simultaneously. And it is difficult to get rid of Gauss noise and impulse noise only by traditional filter at the same time. Based on analyzing traditional filter algorithm, the adaptive-fuzzy algorithm is adopted in this paper. Firstly, the algorithm demarcates impulse noise and separates the two kinds of noise in each part by noise detection. Then the adaptive median filter is applied to remove the impulse noise. Based on filtering the impulse noise, combining histogram and membership function in fuzzy control, a new algorithm is given to filter Gauss noise. Finally there is a brachylogy algorithm for median filter to improve the practicability. In real time application on robots, the new filter method shows better locating performance than traditional algorithm in real time image with Gauss noise and impulse noise.