Reasearch On Surface Flaw Detection For The Sphere And Non-Sphere

Surface flaw is an important indicator of quality evaluation for high precision optical elements, which mainly come from improper operation in the course of manufacturing and using. Operational lifetimes of optical elements and performance of the whole optical system are impaired due to the beam scattering and energy loss caused by the surface flaws. Therefore, it is necessary to detect and estimate the surface flaws of optical elements.At present, studies on surface flaw detection are mainly concentrated on plane optical element at home and abroad. However, these previous detection technologies still existed problems for flaw detection of spherical and non spherical optical elements. In this paper, according to the detection requirements for spherical and non spherical optical elements of the lithography projection optical system, we proposed a surface flaw detection system with microscopy imaging of bright field illumination. Through the analysis of optical systems and mechanical performance index of the system, we discussed the mutual restriction between the optical system imaging resolution and depth of field due to limitation of mechanical motion precision. And on this foundation, we introduced the pupil filter and wavefront coding phase mask into the microscopy imaging system respectively to extend the depth of field. At the same time, we also studied on the key algorithms in the image processing system of flaw detection. The main content of this paper includes:1. Build of surface flaw detection system with microscopy imaging of bright field illumination for sphere and non-sphere surface and analyze its optical systems and mechanical performance index. Through the analysis, we found the key problem of proposed system is the mutual restriction between the optical system imaging resolution and depth of field due to limitation of mechanical motion precision.2. Applications of phase pupil filters with Zernike polynomials and wavefront coding phase masks with non integer power form to extend the depth of field of the microscopy imaging system. We theoretically analyzed their abilityes of depth of field extension and simulated their imagings. These results showed that, for 10 X objective with NA equals 0.2, phase pupil filters with Zernike polynomials could extend its depth of field to 60μm and wavefront coding phase masks with non integer power form was about 100μm, which indicated that wavefront coding phase mask have a better performance. Therefore, we optimized a wavefront coding phase mask further, which could be applied under microscope objectives of different magnification(from 5X to 40X) simultaneously based on the invariance of the defocus modulation transfer function(MTF) of the optical system. The imaging simulations verified its imaging ability of large depth of field for the designed pupil filter. These results showed its effectiveness in the flaw detection system to solve the mutual restriction between imaging resolution and depth of field.3. Study on image processing system of flaw detection system. In this paper, we respectively studied on algorithms in each function units of image processing system of flaw detection system. In the pre-processing unit, an adaptive histogram equalization algorithm was proposed to improve the quality of the original image. In flaw recognition and measurement unit, Canny edge detection algorithm was applied to recognize flaws. In this algorithm, we gave the calculation method of the filter parameters combined with histogram equalizaed results, which ensured the edge positioning accuracy and noise smoothing ability at the same time. In the image stitching unit, we reduced the computation quantity of the SIFT stitching algorithm by employing the mechanical positioning coordinates of the flaw detection system, which improved the operational speed. The time of imaging process for one frame is about 0.8s, which could satisfy the demand of real time detection for flaw detection system.4. Build of verification experiment platform for surface flaw detection of spherical and non spherical. In this paper, we set up the microscopy imaging system with large depth of field by introducing the proposed wavefront coding phase mask, verifying its ability of depth of field extension experimentally. On this basis, we finished the verification experiment of surface flaw detection for a concave spherical lens, whose diameter is 100 mm and R number is 0.8. Experiment results showed that the proposed surface flaw detection with bright field illumination could real-time detect the sphere and non-sphere optical elements of the lithography projection optical system effctively.