基于邻域灰度差值的二维Otsu分割方法研究

2D Otsu method based on difference of gray-level histogram

针对传统二维灰度直方图的阈值分割方法中区域划分像素易丢失、运算速度慢等不足,通过深入分析图像中邻域灰度偏离的情况,并充分考虑像素的空间灰度信息,提出一种利用像素邻域灰度差值的新方法构建二维直方图;基于二维类间方差法实现了图像二维Otsu分割方法,并给出了相应算法的实现步骤。以生物图像中的鱼体分割为实例对方法进行了实验验证,结果表明算法分割效果良好,运算速度提高较明显。

According to the analysis of the threshold method for image segmentation based on the traditional 2D gray-level his- togram which partitioned the areas with some wrong pixels and processes Slowly, this paper brought forword a new method which built the 2D histogram using the difference of the gray-level in the neighborhood. Implemented 2D Otsu method based on be- tween-cluster variance. This method well considered the departure of the gray-level and the information of area gray-level. The experiment shows that the segmentation result of this method is better especial in biological image segmentation and its compu- tation time is reduced much which can apply in the real-time segmentation.