基于数据场和云模型的图像阈值化方法

Image Threshold Method Based on Data Field and Cloud Model

针对图像自动分割中的最优阈值选择问题,提出一种基于数据场和云模型的图像阈值化方法。计算一个与局部相关的全局阈值,根据该阈值将图像自适应划分为若干个子区域,形成四叉树结构描述。通过逆向云发生器生成各个子区域对应的云模型,在云概念空间上产生云数据场,利用数据场的自适应特性实施质心迭代,最终完成不确定性分割,获得图像的二值化结果。该方法从认知物理学的角度重新认识图像阈值的自动优选问题,具有不确定性和空间全局性。分析及实验结果表明,该方法的分割效果较好、性能稳定。

In order to select the optimal threshold for automatic image segmentation, a novel method for image threshold based on data field and cloud model, is proposed. The method calculates the global threshold related with local image. According to the threshold, it adaptively divides the image into several sub-areas, and forms a quad tree. It produces the cloud models for each sub-area using backward cloud generator, generates a data field based on cloud model, and implements centroid iteration using adaptive characteristic of data field. It achieves image segmentation with uncertainty, and yields the binary result. The proposed method recognizes the issue on the automatic optimization of image threshold from the point of view of the cognitive physics, and it has uncertainty and global optimization. It is indicated by the quantitative and qualitative experiments that the proposed method yields accurate and robust result.