Application of polarization imaging in counting plywood layers by machine vision

Abstract

This paper introduces a machine vision system based on polarization imaging, which is applicable for automatically counting the number of internal layers in plywood. Industrial machine vision usually suffers from a low accuracy due to low contrast and high complexity of the images, which could be overcome by the introduction of polarization imaging. A polarization camera was utilized to capture images with polarization angles of 0°, 45°, 90°, and 135°, and then a degree of polarization (DOP) distribution image was obtained by calculating the DOP for each pixel. Compared with the intensity distribution image, the contrast of the DOP distribution image was increased by about 60% and the excessive information in the image including wood’s natural texture, dirty spots, dicing marks, and artifacts was mostly filtered. A gray value difference algorithm was applied to the images to determine the edges of the internal layers of plywood and count them up automatically. The experimental results illustrated that polarization imaging could improve the counting accuracy of the algorithm effectively.