Infrared image edge extraction based on the lateral inhibition network and a new denoising method

The infrared images processing method based on the lateral inhibition network has disadvantages like losing some details of the target or adding noise. Another problem is the parameter of lateral inhibition cannot be chosen adaptively. In the area of infrared guidance, these problems could increase possibility of misrecognition, and lower accuracy of hitting targets. In this paper, we review the theory behind the registration and reconstruction algorithms and their limitations. Because Exponential Function has no parameter to set, this study proposes a method that replaces Gauss Function with Exponential Function to solve the problem of details losing and adaptively image processing. Additionally a method is proposed to remove noise in infrared binary image according to analysis of the neighborhood pixels values of noise point. The results indicate that the method is effective and feasible.