Ear feature region detection based on a combined image segmentation algorithm-KRM

Abstract

Scale Invariant Feature Transform（SIFT）algorithm is widely used for ear feature matching and recognition. However, the application of the algorithm is usually interfered by the non-target areas within the whole image, and the interference would then affect the matching and recognition of ear features. To solve this problem, a combined image segmentation algorithm i.e. <i>KRM</i> was introduced in this paper, As the human ear recognition pretreatment method. Firstly, the target areas of ears were extracted by the <i>KRM</i> algorithm and then SIFT algorithm could be applied to the detection and matching of features. The present <i>KRM</i> algorithm follows three steps: (1)the image was preliminarily segmented into foreground target area and background area by using <i><strong>K</strong></i>-means clustering algorithm; (2)<i><strong>R</strong></i>egion growing method was used to merge the over-segmented areas; (3)<i><strong>M</strong></i>orphology erosion filtering method was applied to obtain the final segmented regions. The experiment results showed that the <i><strong>KRM</strong></i> method could effectively improve the accuracy and robustness of ear feature matching and recognition based on SIFT algorithm.