Aiming to the problem of low computing speed of existing model of fabric characteristic, this paper introduces pre-judgment to improve the speed of image analysis by fabric characteristic model, and constructs gray distribution wave model to quickly estimate image. First, image gray is taken. Then, horizontal and vertical gray distribution wave is established using superposition formula with little computation to further simply model and made the computation speed fast. Finally, from a number of experiments, this model has such features that image can be pre-estimated exactly and continue recognition algorithm can be instructed to leap over some normal regions. So analysis times of image are saved and on-line recognition efficiency of fabric defects is improved.