基于点扩散函数模糊效应的超分辨率图像重构和配准OA

The blurring effect induced by the Point Spread Function(PSF)of an optical imaging system is a core issue that restricts image resolution and measurement accuracy.To break through the diffraction limit and achieve high-precision visual measurement,an integrated solution that combines super-resolution image reconstruction with high-precision image registration is systematically investigated.By establishing a theoretical model based on frequency-domain deconvolution,the blurring effect of the PSF is mathematically characterized,and inverse image restoration is realized.On this basis,an innovative experimental method is designed:leveraging the random Brownian motion characteristics of CdS quantum dots in alcohol,thousands of frames of low-resolution speckle images are collected,which are then used to synthesize a super-resolution digital calibration target image with detail-enhanced features,achieving an equivalent sampling resolution on the order of 0.006 7 pixels.In terms of registration,a calibration algorithm based on homography matrix constraints for the telecentric camera model is proposed.Combined with Harris corner detection and the Bouguet sub-pixel optimization method,the algorithm enables high-precision extraction of feature points and robust solving of external parameters.Finally,horizontal translation and axis rotation experiments are designed to verify the system accuracy.The registration experiments show that under the conditions of this study,the proposed method achieves sub-micron level positioning stability(the average deviations of horizontal displacement and rotation are 0.432 5 μm and 1.146 0 μm,respectively),which significantly improves the resolution and measurement reliability of the imaging system.