关键词: 足迹, 图像分割, 边缘提取, 灰度梯度共生矩阵, 最大熵, 数学形态学

Abstract: The geometrical feature of the edge of human footprint contains lots of human body information, but it is of great significance to extract exactly the edge of footprint. According to the characteristics of footprint image, an approach for extracting footprint edge is presented on the basis of the segmentation of the image area of 2-dimensional gray level-gradient threshold vector. The 2-dimensional threshold vectors can be computed automatically, based on the gray level-gradient cooccurrent matrix model and maximum entropy theory. This method can segment the images by using auto-developed 2-dimensional gray level-gradient threshold vectors, and acquire high anti-noise capacity and high accuracy of segmentation of the fuzzy edge pixies as well. To extract and smooth the footprint edge, 2 structural operators are set up to adopt to the feature of footprint images, with the segmented images computed through a morphologic method. The satisfactory test results show that the footprint edges extracted in this way are of high similarity to original ones with noise suppressed.