Plate Contour Recognized Algorithm Based on Depth Histogram Threshold Segmentation

In this paper, a method of extracting board contour from depth image is proposed, the method has very important practicability in the complex scenes of the current industry, such as uneven illumination and burst noise. The traditional single threshold segmentation methods consider all pixels, the effect of image segmentation in complex scenes is poor and loses more details. Therefore, an algorithm is proposed to automatically select the optimal threshold value according to the features provided by the depth histogram: draw the depth histogram according to the image, the histogram is optimized to reduce the interference information and preserve the image features to the greatest extent. According to the background and the detection target of the frequency distribution in the depth histogram, the optimal threshold can be quickly and accurately determined automatically. The algorithm extracts information from the image depth histogram, optimizes the depth histogram, and eliminates the problem of low robustness of fixed thresholds in complex scenes. The method has been verified on the data set, and the results prove the practicability of the algorithm.