Research on defect detection of the complex texture surface based on the multi-segment computational imaging technique

The defects on the complex texture surfaces can be detected with the traditional machine vision detection technology, which is mainly based on the characteristic differences in the gray scale between the surface intrinsic texture and the defect texture. And also, the height feature can be used in the detection with a 3D sensor. However, the problems can be encountered with these methods, such as the low detection efficiency, high misjudgment ratio, or high cost. To solve these problems, in consideration of the practical application, the defect detection method of the complex texture surfaces based on the multi-segment computational imaging technique (MSCIT) is presented in this paper. On the one hand, the multi-angle images of the surface can be obtained, through the time-sharing trigger of a multi-segment combination of light sources and monocular camera. On the other hand, the gradient information of the surface is restored, according to the different shadow images generated by multiple incident lights and the direction vectors of lights. Then, based on the surface gradient distribution and the image preprocessing technology, the defect information can be enhanced and the intrinsic texture properties of the surface can be weakened. The experiments, carried out on complex texture surfaces with different properties, show that the MSCIT can prevent the useful detected information from being disturbed by the complicated background. The MSCIT proposed in this paper can form a general technology of defect detection of the complex texture surface, and the engineering applications can be achieved.