关键词: 图像分割, 局部阈值分割, 灰度波动, 浮动阈值

Abstract: The industrial image often exists uneven illumination interference, so a local threshold segmentation based on grayscale wave for uneven illumination image is proposed. It extracts image grayscale wave curve in the horizontal and vertical direction, and iteratively searches for larger scale peaks point and troughs point of each curve to meet a given wave amplitude threshold; then, between each pair of alternating peaks or troughs to calculate floating threshold is used to designate the target and background pixels attribution; it takes the intersection of threshold image in two directions to obtain the final segmented image. Experimental results demonstrate that segmentation results and real-time of this method have significantly improved, compared with 2-D Otsu method, 2-D Tsallis entropy method, Niblack method and several algorithms.

Key words: image segmentation, local threshold segmentation, grayscale wave, floating threshold