Detection and Analysis on the Texture-Edges of Machined Surface Images Based on Wavelets

The abrupt point of the signal always carries the important information. In the digital image, the abrupt pixels normally locate at the edge of this image. The texture edge of the image is usually the important characters of the image. The edge detection of texture image is one of the main methods both in the texture analysis and edge detection. The traditional methods of edge detection, such as Gradient operator and Sobel operator are difficult for detecting the real edges and returning many unwanted pseudo edges. So this paper represents a new way for getting the texture-edge of the machined surface images using the wavelet transform. Taking two images (Fig. 1 and Fig. 2) of the machined surface as an example, the results prove that the texture-edge can be abstracted by using the local maximum of Daubechie4 wavelet's coefficients, and the first machined surface, which is produced by the sharp-cut-tool, shows the apparent texture characteristics, but the second machined surface, which is produced by the blunt-cut-tool, shows little texture characteristics.