This paper proposed a new method based on secondary encoding technology to recognize the pattern of woven fabric. The method disposed the information of both the top and bottom fabric images. The result of analyzing the regularity of the intensity variation of the horizontal and vertical direction of image can be used to determine the interlacing position of yarn sections. The secondary encoding technology coded the color information of both sides of fabric image, then the types of cross points can be determined. After getting the color information of fabric on the warp and weft direction, the color pattern could be obtained, and it can be used to correct the errors of classification of cross-point classification. Finally a fabric model can be built which contains types of interlacing points and color pattern.