基于形态学的非均匀光照图像二值化并行方法

BINARIZATION PARALLEL METHOD FOR NON-UNIFORM ILLUMINATION IMAGE BASED ON MORPHOLOGY

在传统工业场合,尤其在大面积视野下的视觉拍摄经常会出现所得图像光照不均匀的问题,这给图像二值化分割以及对后续处理带来困难。非均匀光照图像处理通常采用局部阈值分割法,然而经典的局部阈值分割法通常会具备噪声大、处理时间长等弊端。针对这些弊端,提出利用数学形态学理论,采用改进的Sauvola算法对非均匀光照图像进行二值化研究并对其并行优化。通过对工业场合下采集的卡片图像进行实验验证,结果表明,所述方法不仅可以对噪声进行有效地抑制,得到较好的识别效果,而且图像处理时间也大大缩短。

In tradit ional industrial occasions, especially in the perspective of large area, the captured images often appear non-uniform illumination problems frequently. This brings a lot of difficulty for the division of image binarization and subsequent processing. Non-uniform illumination image processing usually uses the local threshold segmentation method. However, the classical local threshold segmentation method usually has the disadvantages of large noise, long processing time and so on. In order to improve these disadvantages, this paper put forward a method that use the theory of mathematical morphology and the improved Sauvola algorithm for non-uniform illumination image binarization research and parallel optimization. We have experimentally validated the card images collected under industrial conditions. The results show that the method not only can restrain the noise effectively and get good recognition effect, but also shorten the image processing time greatly.