Automatic shoe-pattern boundary extraction by image-processing techniques

A footwear designer digitizes shoe patterns to extract their boundaries, once he finishes 3D shoe-model design, last-bottom flattening, and shoe-pattern making. Shoe-pattern boundaries are then imported to cutting machines to cut material into shoe-pattern shapes. In the shoe-pattern making process, a footwear designer may draw many arcs, lines and pigments on a shoe pattern. Therefore, a shoe pattern has smudges, stains, and marker pen drawings on its surface. It results in the difficulty of automatically digitizing and extracting a shoe-pattern boundary. This study aims to develop an effective image-processing method to automatically extract the boundary of a shoe pattern. In the study, we first use a histogram thresholding technique to segment out a shoe pattern from the scanned input image. Then boundary extraction is applied on the segmented image to detect and smooth the shoe-pattern boundary. Finally, the proposed method is tested and its performance is evaluated. Experimental results indicate that the proposed method is good for automatic shoe-pattern boundary extraction.