Research On Algorithm Of Subpixel Laser Stripe Detection And Feature Extraction

Laser vision detection technology is widely used in medical,archaeological,industrial and other fields because of its advantages,such as high precision,good real-time,non-contact and so on.The high precision extraction of laser stripe center is a very important step in the laser vision measurement system,which directly affects the final detection accuracy.In this paper,the detection and center extraction of different types of laser fringe projection curves are studied.The appearance of steps,bulges and voids on the surface of objects projected by laser will lead to the lack of extracted information.Some objects strongly reflect light or absorb light,which makes the effect of the same laser stripe different.These problems will bring difficulties to the extraction of the laser stripe center.This paper analyzes the laser stripe image;several laser fringe center extraction algorithms commonly used are summarized.And it proposes an improved method of normal centroid method for extracting the center features of laser stripes,which accuracy can reach sub-pixel level.The difference shadow method and the connected domain method are used to segment the stripe image twice,and then the median filter is used to reduce the random noise,so that the effective filtering of all kinds of noise can be realized.In the direction of the center feature extraction,the centroid method is improved.Firstly,the filtered stripe image is etched and refined to obtain a stripe skeleton.The grayscale gradients values are calculated for each skeleton point based on the neighborhood point of the skeleton to obtain a stripe normal.Then the nearest light points to the normal are found,and the sub pixel center coordinates of the stripes are extracted by the centroid method.Experiments and simulation results of laser stripe on different projection surfaces show that the algorithm has high extraction accuracy and speed.Finally,the sub-pixel laser stripe detection and center extraction software system is designed and implemented in order to facilitate the application of the algorithm,and the simulation analysis of weld inspection is carried out.The results show that this method has the advantages of high extraction accuracy,wide practicability,small computation and good robustness.