基于摩尔纹光斑的镜头模组调焦算法研究OA

Aiming at the problems of low efficiency and insufficient precision in the production process of fixed-focus lens modules,a real-time high-precision Moiré focusing method based on vision is proposed.By adopting a collaborative optimization strategy that integrates local threshold segmentation and adaptive threshold algorithms,the problems of eccentricity and optical interference in spot images of different lens modules are solved.The local threshold segmentation algorithm achieves sub-pixel-level precise alignment of Moiré pattern spots,enhancing the adaptability to Moiré pattern feature extraction for different types of lens modules,and the recognition rate can reach 99%.Real-time image processing based on zero-copy frame capture,combined with an intelligent standard deviation pre-screening mechanism,achieves an optimized allocation of computing resources,improves the detection efficiency of the lens module,and reduces the single-frame processing time to 24 ms and the focusing time of a single lens module to 8 s.The dynamic threshold segmentation model can maintain a stable detection accuracy of over 98%even under nonlinear and non-uniform illumination conditions.Compared with the traditional Moiré pattern detection methods,the algorithm increases the detection efficiency by 40%and the recognition accuracy by 18%.