| Summary: | 碩士 === 國立臺灣科技大學 === 高分子系 === 96 === Enabling computer vision to recognize texture of different fabrics is an important factor in fabric analysis and a progress in research. Throughout the process, the warp and weft floating points of the fabric’s image determine the fabric’s texture. However, different optics environment, fabric materials, and computer vision method stability can lead to variable results, causing computer vision system’s usability and the fault-tolerant ability to be focused in research. We proposed a new automatic recognition algorithm for fabric weave pattern recognition. We also used neural network to construct the computer vision system to recognize fabric’s texture, and to increase this system’s reliability and fault-tolerance. In this study, we used first- and second-order statistics method and composed the classified system with two-step back-propagation network. Experimental results indicated that fabric patterns can be identified clearly by our proposed method.
|
|---|
