Study And Realization Of Image Processing System For Solid Residue And Defect On Surface Of KDP Crystal

KDP(2 4KH PO) crystal, because of its excellent optical performance, easily growing a large-sized single crystal and other characteristics, is widely used in ICF(Inertial Confinement Fusion) system. With the development of science and technology, ICF system requires higher quality of optical element surface. There would be surface defects frequently on crystal during single point diamond turning and MRF(Magnetorheological Finishing) polishing, such as solid residue, scratche, cutting corrugated texture,which may seriously degrade the optical properties of the KDP crystal. The traditional and qualitative surface quality detection methods, however, can not quantify the surface damage after KDP crystal processing.On the basis of lots of literature about the quantitative quality detection methods used on optical element surface, optical microscope is used to obtain the surface image and image processing system is designed to achieve quantitative detection of defects on surface of KDP crystal in the paper. The effort is important to improve KDP crystal surface quality. The main contents and conclusions of the paper are as follows:1. In the paper, image samples of KDP crystal surface after processing are acquired, and the image pretreatment methods including gray-scale processing, image filter and uneven illumination correction are used. Through identifying the image scale, the calibration coefficient of image is obtained.2. Extracting features and removal of cutting corrugated texture is showed. Through the analysis and comparison of calculation method for the main direction of texture, which is based on maximum curve of column means and Hough Transform, the paper presents a conclusion that the latter is much more accurate and efficient. The main grain direction, image rotation and gray projection algorithm are introduced to characterize the column gray value curve which the image is in the vertical direction. Temporal filtering is used to obtain the curve’s peaks and valleys, and in particularly, calculate the average spacing of texture.Based on the main grain direction and Fourier Transform, the frequency-domain filter that is used to remove the texture is constructed. For analyzing the relation between the characteristic parameters of GLCM(Gray Level Co-occurrence Matrix) and ?? which is the key parameter to remove texture based on Fourier Transform, the influence of GLCM growth factor is discussed, including growth step, gray scale and growth direction. At a result, ASM(Angular Second Moment), CON(Contrast), IDM(Inverse Difference Moment) and ENT(Entropy) show the regularity that there is a relationship between the parameters and increasing of ?? or removaling of texture.3. The recognition technology of the solid residues and scratches based on the shape features is researched. Two-dimensional Renyi entropy and morphological processing are used to realize image binarization, contour modification and edge detection. And experimental analysis of the impact for key parameters α on the two-dimensional Renyi entropy segmentation shows its optimal range is between 1.5 and 2.5.Freeman chain code is introduced to calculate edge curvature. The threshold of curvature and line criterion are used to estimate the scratche boundary. On the basis of shape features, the solid residues and scratches are accurately classified.4. Eight evaluation parameters are extracted about the solid residue, scratche and cutting corrugated texture. The image detection and recognition system of KDP crystal surface has been improved.