A monocular vision-based recognition and location algorithm of motor pedestal part for hair dryer assembly line

To improve the assembly efficiency of hair dryer, an automatic detection system based on monocular vision is applied to identify and locate motor pedestal parts for hair dryer assembly line. Due to the motor pedestal parts usually have similar shape and low discriminating features, it is difficult to recognize and locate them with high accuracy. In this paper, a recognition and location algorithm is proposed to identify and locate multiple motor pedestal parts simultaneously for robotic grasping during assembly phase. First, the original images are filtered and divided into many sub-images that every sub-image only contains a single motor pedestal part. Secondly, a masking operation is performed for a motor pedestal part in every sub-image to extract contours of the part and the type of the part is identified according to the number of the contours extracted. Subsequently, the sub-pixel edge of the part is detected by using an improved Devernay algorithm and then the location of the center of the part is determined. Finally, the actual location coordinates of the parts are obtained according to camera calibration. Type recognition and location detection experiments were carried out and the experimental results demonstrated that the proposed algorithm had high accuracy and good suitability.