Aiming at overcoming the influence of the image noise to the image segmentation performance, This paper extracted the non local spatial information of the pixel from a set of pixels with a similar neighborhood configuration of the current pixel, and then constructed a two-dimensional histogram based on the gray and non-local spatial gray information. This novel histogram was introduced into the Otsu' s curve thresholding segmentation algorithm and it proposed a two-dimensional Otsu' s curve thresholding segmentation method based on gray and non local spatial gray feature. Segmentation experiments show that the proposed method further improves the robustness of traditional two-dimensional Otsu' s curve thresholding method to image noise and obtains more perfect image segmentation results.%为了克服图像噪声对图像分割结果的影响,利用图像中与像素具有相似邻域结构的像素提取当前像素的非局部空间信息,构造了基于像素的灰度信息和非局部空间灰度信息的二维直方图,并将此二维直方图引入到Otsu曲线阈值分割法中,提出了基于灰度和非局部空间灰度特征的二维Otsu曲线阈值分割法.实验结果表明,该方法能进一步提高原始二维Otsu曲线阈值分割法对于图像噪声的鲁棒性,获得了更加理想的分割结果.
展开▼