Super‐Resolution Microscopy for Precision Microsphere Defect Inspection Using Sparrow‐Optimized Autocorrelation Deconvolution

Precision microspheres have small volumes, making the detection of surface defects challenging with the naked eye. Traditional optical microscopy methods are hindered by issues such as localized blurriness and low resolution, which impede their ability to detect surface defects of microspheres with the precision required. This paper proposes an autocorrelation two-step deconvolution super-resolution image reconstruction method using the sparrow search algorithm, which adaptively fine-tunes the acceleration parameters and optimizes them through a hybrid energy function as the objective. This method enhances image resolution and mitigates the occurrence of artifacts during the reconstruction process. Rolling Fourier ring correlation and full width at half maximum are employed to assess the quality of the reconstructed images. In comparison to the traditional autocorrelation two-step deconvolution super-resolution algorithm, the proposed method achieves a resolution enhancement ranging from 2.79 to 3.82 times, depending on the frame count. For equivalent frame counts, the image processing speed increases by 6.3%-34.2%. When compared to various deep learning models, the proposed algorithm reconstructs more detailed features and enhances the detection of surface defects in precision microspheres. It displays a depression effect, which may serve as a valuable reference for the quantitative assessment of surface roughness in precision microsphere images.