Quantitative assessment of laser speckle uniformity via Tamura coarseness analysis: algorithm development and experimental validation.

Laser speckle patterns generated by multimode optical fibers pose significant challenges for uniform illumination, while existing evaluation methods often fail to provide reliable quantitative assessment due to their limited ability to capture the spatial and perceptual characteristics of speckle fields. In this work, a perception-consistent framework, termed Tamura coarseness-based laser uniformity quantification (TC-LUQ), is proposed for quantitative evaluation of laser speckle uniformity. The method integrates multi-stage image preprocessing with multi-scale Tamura texture analysis to establish a robust relationship between visual perception and objective uniformity metrics. A dynamic tri-partition histogram equalization and an improved non-local means denoising scheme are employed to enhance contrast and suppress noise while preserving speckle structure, followed by a multi-scale coarseness model for quantitative characterization. Experimental validation using a mechanically controlled homogenization system demonstrates that TC-LUQ produces evaluation results consistent with visual perception. Under optimized conditions, the experimental homogenization system achieves a minimum roughness value. These results indicate that TC-LUQ provides a reliable approach for laser speckle uniformity evaluation and supports the optimization of laser homogenization systems.