Characteristic Feature Extraction In Complex Background

Taking the tread in alibi as object, this thesis introduces some knowledge about image processing and pattern recognition, and extracts characteristic feature in complex background. Tread feature recognition is one of the most important biometrics in criminal investigation work. A scheme was developed to correctly reproduce distinct, continuous edges and decrease manual intervene based on the maximum variance between clusters (Otsu) method and the algorithms of fuzzy C-means clustering. In this paper, an improved Otsu algorithm is proposed, which is based on the two-dimensional bound histogram. First the Otsu method and the algorithms of fuzzy C-means clustering is used to segment the heavy pressure surface form the image, then the morphological filter and area threshold removing method are applied to filter the small area and the noise, with an adaptive method to select the area extraction threshold. Experimental results show that the scheme reproduces accurate, smooth edges due to the use of the gradient and gray information, providing a new way for tread feature automatic recognition.