文件名称：image feature
文件大小：883KB
文件格式：PDF
更新时间：2013-01-04 15:05:04
image feature,representation
For the two-dimensional object silhouettes, in this study, we present a one-dimensional descriptor which in theory remains absolutely invariant under affine transforms. The proposed descriptor operates on the affine enclosed area. We design a contour normalizing method. After this normalization, the number of points on a contour between two appointed positions remains unchanged by affine transforms. We prove that for any linearly filtered contour, the area of a triangle whose vertices are the centroid of the contour and a pair of successive points on the normalized contour remains linear under affine transforms. Experimental results indicate that the proposed method is unaffected by: boundary starting point variations and affine transforms even in the case of high deformations and serious noise on the shapes. We also propose a method to simulate the noise contaminating the test shapes, in a real context, and define the signal-to-noise ratio for a shape. In addition, the proposed normalization method can be associated with other algorithms to increase their robustness to affine transforms and decrease their complexity in similarity measures.