Research Of Fabric Defect Detection Algorithm Based On Back Projection

Defect detection is an important means to ensure the quality of textile and an important link of the textile production process. At present, most of the domestic fabric detection still relies on artificial eye detection which has high labor cost and low efficiency. While, the detection method based on machine vision which using automated image acquisition and defect recognition can greatly make up for the shortage of artificial detection and is gradually replacing artificial detection to become the future development trend. Thereinto, the image processing algorithm is the focus of research on fabric defect detection.According to the periodic texture of fabric, common texture feature extraction methods for fabric defect detection mainly include statistical method, signal processing method, structural method and model method. Although these methods have some effects, they generally have high computational complexity and take long time. So this paper analyzes the fabric defect detection algorithms based on back projection. The main research contents are as follows.Based on histogram back projection, the detection algorithm based on gray level histogram back projection is studied. The gray level histogram back projection is improved, the peculiarity of texture shielding is introduced, and the detection process and parameter selection of fabric defect detection are analyzed. Then, experimental results are given. Finally, the test results and computation time are compared.According to the problem of bulk oil defect detection in gray level histogram back projection, the improved gray level histogram back projection detection algorithm is proposed. Three ways including the detection process, the block feature and the back projection principle are analyzed. Finally, the improvement of back projection and subsequent detection process are adopted and the experimental results are given.According to the unicity of gray histogram characteristic, the detection algorithm based on gray level co-occurrence matrix back projection is proposed based on gray level co-occurrence matrix and improved back projection. The principle of the algorithm is introduced, and the process and parameter selection of the algorithm are analyzed, then the experimental results and the comparison with other methods are given.In summary, through the test of above three fabric defect detection algorithms based on back projection, the following conclusions are obtained. Improved gray level histogram back projection detection algorithm and the detection algorithm based on gray level co-occurrence matrix back projection are the improvement of the detection algorithm based on gray level histogram back projection and have better detection effects. Defect segmentation effect of the detection algorithm based on gray level co-occurrence matrix back projection is better, but its computation time is long than the other two. If pursuing the robustness of algorithm, the gray level co-occurrence matrix back projection should be selected. If pursuing the detection speed and accuracy, the improved gray level histogram back projection should be selected.