Study Of The Stitch And Fusion For The Holographic Images Of Precision Parts

Holographic interferometry technique has been widely applied in the field of precision measurement. The digital holographic technology compared with traditional holographic technology, broke the restrictions of the traditional holographic on the external environment and film,and its reappearance characteristic has the incomparable advantage over the traditional holographic technology, it also increased the phase analysis of data, make it has wide foreground in the field of measurement. The intention of the thesis is to solve the crucial problem of the digital holography, it has made the analysis and research, and take the corresponding solutions.And the traditional holography and digital holography has been taken a different approach, the problem in the field of holographic visual was overcome in this thesis.In this thesis, the relevant research situations at home and abroad were analyzed, the principle of Holographic imaging was researched. Different kinds of noises between the traditional holographic photos and the digital holographic images are analyzed. The corresponding denoising scheme to filter the noise based on the holographic image of time domain and frequency domain was proposed. In this way, the zero-order image and the periodic stray light were erased, so the quality of the the holograms and the reconstruction images was improved. The CCD size limits the scope of holographic images, affecting the measurements of the precision parts. To slove the problem, the holographic images must be stitched. As the resolution and quality of the hologram are different between the traditional holography and the the digital holography, the multi-scale Harris detector and SIFT detector were taken respectively and the L-M algorithm based on the RANSAC algorithm was proposed to obtain the homography with high accuracy. To extract the holographic image from a wide range of image background, the holographic image segmentation is indespensiable to get effective information in an image hologram area and make a separate analysis, this thesis aimed at different hologram design different segmentation method to meet the requirements of image fusion and the analysisof phase information. The improved watershed algorithm was proposed to prevent the transition segment phenomena caused by the traditional segmentation method. As high connectivity of the traditional holographic representation images, the FCM algorithm based on clustering segmentation method was adopted to realize the accurate segmentation of traditional holographic representation. For the resolution and sizes of the representation images with different representation distance and representation angles are different, the improved pyramid fusion algorithm that different matching rate with different rules was proposed to realize the traditional holographic image integration. Another algorithm based on compressed sensing technology was proposed to slove the problem of large data fusion of low efficiency for digital holography.In order to validate algorithm that included Denoising, splicing, division and integration proposed in this thesis, the digital holographic measurement device of the precision machinery parts, the experinments as follows:(1) The roughness of the standard sine roughness specimens was measured five times, the experimental results show that the maximum deviation of the Ra values measured by digital holographic measurement and the nominal Ra value is 0.00182μm.(2) 36 digital hologram was token for the precision machinery part, the whole data of the hologram reconstruction was obtained after the process of the denoising, splicing, separation and fusion, the sizes of the six pores on the machinery part were measured.it shows that the maximum machining error of the apertures is 0.65%, the the roughness of the standard deviation is 0.0017μm.