Detection of packing filter rods based on visual saliency
Zhou Shuaijun, Dong Ning, Jun Du, Sai Yang, Zhi Xiang Yang · 2016
Concerning the diversity of packing filter rods' defect type and low efficiency of traditional artificial detection methods, in order to detect the defect more effectively, an approach to detect packing filter rods based on visual saliency was proposed. The images of packing filter rods were utilized to be clipped and high quality saliency cut, then to extract various effective saliency features to compute the salient value. The different types of saliency maps was fused to final saliency maps to finish saliency detection. Four classical saliency models are used to compute the salient value. And these saliency maps was compared in the image test database built by us. According to the result, AC has higher precision and better recall.