Dynamic Recognition of Fabric Wrinkle from Image

Abstract

An improved genetic algorithm is employed in recognizing and detecting the fabric wrinkle deformation. Firstly, the 3D features of the fabric wrinkle may be captured before it moves along running direction continuously under the internal and external forces. Then the feature area is probably localized and finitely discredited to build a feature library. When the deformation occurs, the images may be captured and be also represented by the discrete points, which can be defined as the initial population. Its genes may be represented by the color values and positions of all discrete points. Secondly, the most similar discrete points, found out by comparing all the discrete points and those of the feature library, are taken as the revolutionary direction. In addition, the snap angle may also be calculated among the revolution. The experimental results show that such algorithm is capable of efficiently detecting the fabric wrinkle deformation.