基于OpenCV的复杂环境下图像二值化方法

Binarization method for images acquired under complex environment based on OpenCV

机械式电表读数图像易受光照不均、反光等因素影响,导致传统二值化方法效果不佳。针对以上问题,基于OpenCV开源视觉库,提出一种结合改进的Canny边缘检测算法、全局阈值与局部自适应阈值的图像二值化方法。从图像自身特点出发选取Canny算子最佳全局双阈值,并进行局部动态阈值修正实现准确的边缘检测。综合图像边缘信息、全局阈值与局部自适应阈值实现图像二值化。实验表明在光照不均、反光等因素影响下,该方法取得了良好的二值化效果。

The mechanical meter numeral images often acquired under non-uniform illumination and reflection.The traditional methods can＇t binarize them correctly.To solve the problem,a binarization method based on OpenCV by combining Canny＇s edge detection with global threshold and local adaptive threshold is proposed.In order to improve the edge detection effect,the high and low threshold parameters for the Canny operator is determined according to the characteristics of the meter numeral images,and the local dynamic threshold is involved for correction.The binarization method is achieved by the fusion of the image edge,global threshold and local adaptive threshold.The test shows that the method can achieve good effect under non-uniform illumination and reflection.