Abstract: During the actual image has existing object and background gray level information distribution imbalance, based on the image thresholding thinking focused on the goal, the histogram information of the image is introduced to get more detailed guidelines thresholding. Considering the edge information of the image, the gray gradient mapping function is introduced, the gradient based and the focused on the goal of Otsu threshold method is proposed. The results of lots of classic image thresholding show that the proposed method is better than the existing methods, and it can obtain more complete objects and clearer edges.

Key words: image segmentation, focused on the goal of Otsu, gray gradient mapping function

摘要： 针对目标与背景灰度分布不均匀的图像，基于集中于目标的图像阈值法思想，引入图像的灰度直方图信息，得到更为细致的阈值化准则。考虑图像的边缘信息，引入灰度梯度映射函数，提出了基于梯度的集中于目标的Otsu阈值法。大量经典图像阈值化结果表明，该方法在目标提取的完整性和边缘保留的清晰性方面，均表现出了更佳的效果。

关键词: 图像分割, 集中于目标的Otsu法, 灰度梯度映射函数