Off-line handwritten signature verification with inflections feature

In this paper, an off-line handwritten signature verification system is proposed. Different scale wavelet transforms are used in the curvature signature signals transformation. After analysis, proper scale is selected. We extract the zero-crossings points and take it as the inflections of the signature. Then the signature curves are divided into several parts, i.e. the strokes, according to these inflections. The distance between two corresponding strokes can be measured with dynamic time warping algorithm. In the end, the training algorithm of the signature verification system and the verification method of the signatures are also introduced. Experimental results verify the effectiveness of this method.