一种岩石薄片离焦模糊图像自动筛选方法

An automatic screening method for defocused fuzzy image of rock slice

针对岩石薄片全幅面偏光采集中,部分视域采集的图像会存在离焦模糊问题。本文提出一种模糊图像检测方法,能在薄片全幅面采集的大量偏光图像中,自动检测出存在离焦模糊的视域。由于图像模糊会导致图像空间域和频率域的一些特征产生变化,因此本文结合空间域和频率域方法,对图像分块进行模糊评价得到模糊度图,由该图统计特性将图像区分为清晰图像、局部模糊图像或全局模糊图像,同时还验证了可以将该模糊度图直接用于图像模糊区域分割。实验结果表明,本文算法可以有效筛选出存在模糊的视域,并且在模糊区域分割上也具有较高的准确性。

In the full-width polarizing light acquisition of rock slices,some view fields may be suffered from defocused blur.In this paper,we propose an effective approach for detecting blur image in a large number of polarizing images collected in the whole width of the slice automatically.As image blurring will lead to some characteristic changes on both of the spatial domain and the frequency domain,we combine them for image blur estimation based on the image block to produce a blur map.The image can be classified into clear image,local blur image or global blur image by the statistical characteristics of the blur map.By the way,it is verified that the blur map can be directly used for image blur segmentation.Experimental results show that the proposed algorithm can effectively screen out blur view areas,i.e.,local blur images or global blur images,and has high accuracy in blur region segmentation.