Research and Development on Automatic Color Texture Identification and Pattern Making Process for Embroidery Fabrics

99

Currently, there are very few literatures on automatic image recognition and classification of embroidery fabrics. In today’s embroidery industry, front-end pattern-making still relies greatly on labor, using pattern-making software to carefully depict patterns and images in different colors and regions. Hence, an image analysis system that can recognize colors, regions and patterns automatically is a critical technique of improving the competitiveness of the embroidery industry. In this dissertation, the mean filtering method, central weighted median (CWM) filtering method and morphological operation are employed to filter out the light variation on embroidery fabric surface structure, and apply Genetic Algorithm (GA) to distinguish images of repeat pattern embroidery from that of non-repeat pattern embroidery. If it is a repeat pattern, then a much smaller sized sub image would be searched in the original image for the same color components and spatial structure, which could lower the computing load of the entire image greatly and is expected to achieve the processing speed required in an online real-time system. As for non-repeat pattern embroidery images, discrete wavelet transform (DWT) is applied to acquire low-frequency sub images, which can retain important image features while improving the computing efficiency. Be it a repeat or non-repeat pattern, after obtaining sub images, Specific Criteria (SC) is used to determine the exact number of clustering, and the weighted fuzzy C-means method (WFCM) is employed to run color separation and region separation. The experiment proved that, in regard to the color embroidery images of repeat and non-repeat patterns, the method proposed in this dissertation succeeded in color and region separations with good result. The embroidery fabric is different from other planar fabrics such as printed fabrics and twill fabrics. Because embroidery fabrics have inherent solid texture patterns, furry edges, voids and thickness shadows, it is very difficult to filter and simulate texture patterns and is the bottleneck for embroidery automation. In view of this, this dissertation proposes the texture fitting method (TFM). TFM is a kind of non-filtered digital image processing method. For embroidery fabrics full of multiple single-connected, single-color and single-texture closed regions, TFM can fast complete color and region separation and texture simulation, and then output the result to monitors or plotters to investigate the simulation effect and compare it to real fabrics, or use this technology as a generalized filter for embroidery fabrics. This dissertation first addresses a combination of mean, morphological and central weighted median filters to remove light variation on embroidery surface, periodic darkness on the greige and noised texture structures so as to separate colors by WFCM and reshape 1D image pixels to finish region separation. The second part of this dissertation utilizes TFM to identify stitch colors and simulate texture patterns over the whole image. By exporting the result to visual devices, we can prove the integral correctness and efficiency of the texture simulation.