An Accurate Detection System of Foreign Fibers in Cotton

摘要

The types of cotton foreign fibers are diverse and complex accordingly, it is difficult to achieve an accurate segmentation. This paper proposes a superimposed segmentation algorithm based on Spatial Domain Image Enhancement and S-channel Enhancement. One thousand samples are processed in experiment: on the one hand, gray images are achieved after cotton foreign fibers samples are dealt with gradient processing from north direction and threshold adjustment; on the other hand, other gray images are achieved after cotton foreign fibers samples are dealt with extraction and enhancement based on S channels. Experiment show that these two kinds of enhancement methods have complementary advantages. Finally combining the gray images processed respectively by these two methods can separate the target and the background effectively and achieve the accurate segmentation of cotton foreign fibers.