Research Of High Precision Derection Algorithm On The Precision Optics Lens Surface Scratches Based On Image Processing

With the rapid development of optical technology, precision optics have been widely used in military, industrial and other fields, and the quality requirements of the optical surface of the lens has been rising. Optical lens in the process will inevitably produce some defects (such as scratches, broken edges, spots, etc.). These defects will cause diffraction and scattering and seriously affect the performance of the optical system especially cause by scratches. Therefore, to achieve the optical lens surface flaws particularly fast and accurate automated detection of scratches has always been a difficult problem to be solved.This article mainly focused on solving the problems in detecting scratch flaws on the optical lens, such as low accuracy and poor robustness. On the basis of the existing algorithms, a new scratch flaw detection algorithm which was based on the multi-scale line detection combined with the sub-pixel edge detection algorithm on the basis of discrete orthogonal polynomials Gaussian curve fitting, its main contents are as follows:(1) As the interference factors in the optical lens’ surface images, such as lots noise, low contrast and uneven illumination, etc, which led to scratch detection effect was poorer, an improved multi-scale line detection algorithm was put forward. Depending on the different response characteristics, the multi-scale weight was improved in this algorithm, which was making full use of the detector with different scales. Experimental results showed that, this algorithm had some advantages when compared with the traditional edge detection algorithms, such as improving the processing effect of scratches in a certain extent, lessening the affected by noise and uneven illumination, and being not sensitive to the selection of threshold value. At the same time, this method is also applicable to detection of general type roof edge.(2)To further improve the detection accuracy scratch flaws, based on the researching of existing sub-pixel edge location algorithm and summarizing the advantages and disadvantages as well as the scope of application of different algorithms, and combined with the research object, an improved sub-pixel edge detection algorithm was proposed which was based on discrete orthogonal polynomial curve fitting. Firstly, select the discrete points in the gradient direction of the edge by the edge points. Second, fit the selected discrete points into curve by the discrete orthogonal polynomial and compute the coordinates of sub-pixel edge point by the fitting curve. Experimental results showed that the proposed method had obvious advantages in terms of accuracy and efficiency of segmentation, with better usability and robustness.