Based Neural Network Adaptive Thresholding Method Parameters Automatically Determine Its Application In Color Image Segmentation

In this paper, two methods are first introduced that they aim at improving on LEGION (Locally Excitatory Globally Inhibition Oscillator Network).First, the function for evaluating image segmentation techniques are introduced into LEGION, and which accomplishes automatic setting parameters of LEGION. Second, it creates a three-dimensional method of LEGION for segmenting color image. Every object and important details of image are segmented distinctly. Finally, the compound arithmetic of LEGION is created, and it is good for segmenting color images that include texture and notexture.The creative points of this paper lie in the following:1 It is first that it applies evaluating function to LEGION, and which advances the automation of LEGION.2 It creates a three-dimensional arithmetic of LEGION for segmenting color image.