Knowledge-based automatic extraction of multi-structured light stripes

To achieve automatic processing of the multi-structured light stripe images, we need to extract many stripes automatically in a short time. However, due to the complexity and diversity of its work, the related research progress is very slow. Therefore, we first use the Radon transformation and grayscale transform enhancement to eliminate noise in images. Then, a new and adaptive feature model is created based on the knowledge of stripes distribution. The distribution of the stripes is matched by the model, and then the target stripe regions are picked up to realize the extraction of the stripes. In the actual verification, the qualified rate of the detection is basically over 80%, and the detection time is controlled at about 2 s. The automatic extraction of target stripes effectively avoids the tedious work that workers need to do it manually, and it is of great significance for the application of multi-structured light detection technology.