基于纹理分析的小波变换图像清晰度评价方法研究

Research on image definition criterion using wavelet transform based on the texture analysis

验证分析了小波变换在图像清晰度评价上应用的可行性,研究发现基于小波变换的图像清晰度评价函数性能与图像的空间纹理分布有很大的关系,采用图像灰度梯度方向矩阵的纹理分析方法,可以显著提升其性能。由于机加工零件表面纹理的普遍存在,对此类图像进行纹理分析有了实际应用意义。在对图像纹理特征进行统计分析的基础上,本文提出了一种基于纹理分析的图像小波变换清晰度评价方法。经实验验证,此评价函数针对具有纹理特征的目标有优越的性能,具有较高的灵敏度和分辨力(优于2μm)。

The feasibility of wavelet transform in image definition criterion application was verified and analyzed. Research shows that the performance of image definition evaluation function is significantly related to the spatial texture distribution of the image, and image gray gradient direction matrix texture analysis method can greatly improve its performance. Due to the prevalence of surface texture on machined parts, the image texture analysis has gained practical applications. Through statistical analysis of the image texture characteristic, an image definition criterion using wavelet transform based on the texture analysis is proposed in this paper. Experiments show that the evaluation function has superior performance for objects with texture features, and achieves high resolution （better than 2 μm）.