Cable Target Image Segmentation Method for Manual Cable Laying Process

The cable bending curvature in manual cable laying is one of the key performance indicators,which exerts influence on cable laying quality.In order to obtain cable target area being assembled in cable laying process,a target image segmentation method considering manual operation is proposed to get the cable region image during the operating.Given a cable image from egocentric vision,all minimum rectangular bounding boxes for cable target regions are calculated according to cable connected areas segmented in HSV color space.Meanwhile,hand areas in cable assembly image are extracted in YCrCb color space and thumb points are detected by curvature analysis method.Then,the positional relations between thumb points and all rectangular bounding boxes are estimated on the basis of 2D orientation bounding box intersection principal.In this way,the target bounding box belonging to cable target region can be found.Finally,contour of cable target image is pinpointed by target bounding box,and it helps to get the cable target image by producing mask image with flood-fill algorithm.The experiment shows the method above could meet the requirement of segmenting the cable target image,which can be used to analyze curvature of assembling cable.