Multi-scale Segmentation Method for Brake Beam of Freight Train

A segmentation method comprising gray projection, discrete wavelet transform and adaptive segmentation was proposed to realize automatic failure recognition for trouble of moving freight car detection system (TFDS). The correlation between distribution of gray scales and their location were considered when segmented the objects. Calculate the vertical gray projection. Fit the centerline accordingly based on the extreme points which represented the edges of break beam. Considering priori information in the image, rotate a rectangular area in the plane and calculate gray projection along the long axis within the area. Analyze the projection curves and the inclined brake beams could be segmented. Experimental results illustrated that, the technique performed excellently for images which were underexposed, overexposed, local reflex, and blurred.