通过对参数空间的θ参数进行低、高分辨率分级, 用迭代的方法, 由粗到细, 逐步细化参数空间的θ参数, 直至达到要求的精度. 该算法先后需要进行两次Hough变换：低分辨率全局Hough变换和高分辨率局部Hough变换. 研究了当θ参数的分辨率△θ0在2°、3°、4°、5°、6°、7.5°、9°和10°不同取值时, 多分辨率Hough变换对焊缝图像的分析结果, 并与同等精度下的标准Hough变换分析的结果进行了对比. 结果表明：本文提出的多分辨率Hough变换与标准Hough变换分析的结果一致, 且内存需求降低, 运算速度得到提高。当△θ0为6°时, 运算速度至少提高三倍。

The parameter θ of parameter space was classified as different resolutions and the θ axis of parameter space was refined to required precision by iteration method. Two different Hough transformations shall be operated: Low-resolution Globe Hough Transformation and High-resolution Regional Hough Transformation. Under different of parameter θ (△θ0) value: 2°, 3°, 4°, 5°, 7.5°, 9° and 10°, analysis results of Multi-resolution Hough Transformation (MHT) Analysis of Welding Image resolution were researched and compared with Standard Hough Transformation (SHT). It demonstrates that the results of MHT coincide with SHT, moreover, the memory requirements decrease and calculating speed increases three times when the value of △θ0 is 6°.