Method for extracting and recognizing human ear characteristic by improved Hausdorff distance

Abstract

The invention relates to a method for extracting and recognizing human ear features by using improved Hausdorff distance. The invention obtains the standard human ear image through the preprocessing of the human ear image, including the collection of the human ear image, the denoising of the non-skin noise, the normalization of the space size and the illumination compensation. Then, the feature of human ear edge is extracted by gray-scale morphological gradient and local threshold segmentation method, and the standard human ear edge image is obtained. By adopting the method of the present invention, the improved Hausdorff distance through the standard deviation and the length difference between the edge line segments reduces the influence of point set non-contour edge line segment points (outfield points), obtains better anti-noise performance, and enhances the Hausdorff distance obtained based on the Hausdorff distance. The eigenvalues ​​are used for the accuracy of the recognition of the edge image of the human ear, which greatly improves the recognition rate of the human ear.

Description

technical field [0001] The invention belongs to the personal identification technology based on human biological features, in particular to a method for extracting and identifying human ear features by using improved Hausdorff distance. technical background [0002] Human ear recognition technology is a biometric recognition technology that began to emerge in the late 1990s. The unique physiological characteristics of the human ear and the advantages of observation angles make the human ear recognition technology have considerable theoretical research value and practical application prospects. The human outer ear is divided into the auricle and the external auditory canal. The object recognized by the human ear is actually the exposed auricle of the outer ear, which is what people are used to call the "ear". A complete automatic ear recognition system generally includes the following processes: ear image acquisition, image preprocessing, ear image segmentation, feature extr...

AI Technical Summary

Problems solved by technology

However, since the Hausdorff distance is a maximum and minimum distance, it is still very sensitive to the noise in the human ear image

In human ear recognition, although there may be two very similar or even identical human ear edge images, due to the existence of some false contours (non-human ear main contour segments), the calculation of the Hausdorff distance will also produce a large error.

Moreover, edge detection itself is an uncertain problem. The definition of edges is usually very vague, and it is very difficult to distinguish edge points from noise points.

Images