基于知识粒度的图像分割新算法

New image segmentation algorithm based on knowledge granularity

为有效解决粗糙理论在边界域不能够变化因而无法适应图像信息复杂空间的相关性和不确定性的问题,提出基于变精度分层粒度模型的图像分割算法。以知识粒度为基础,引入分类误差精度,构造出具有不同置信阈值和分类质量的图像粒度结构;根据分割精度要求,确定单元粒度层,在该粒度层分析不同灰度级的重要度,进行相应的灰度核计算;通过差异度指数定义等价关系,实现相似区域合并,完成图像分割。分割实验结果表明,该算法降低了图像信息和时间的复杂度,提高了图像分割的并行性,为知识粒度在图像处理中的应用提供了新思路。

To solve the problem that the rough theory in the edge boundaries is not able to adapt the correlation of image infor-mation complex space and the defect of uncertainty ,a new image segmentation algorithm was proposed based on the variable pre-cision hierarchy granular model .Firstly ,the classifying error precision was introduced into the knowledge granulation ,and the granular structure of the image with various levels of confidence and the classification quality was constructed .Next ,on the basis of the requirement of the segmentation precision ,the unit granular layer was chosen and the importance of different grey levels on the layer was analyzed further .Finally ,the equivalent relations defined by using the dissimilarities were used to implement the combination of the similar regions and the image segmentation was accomplished .The algorithm was applied to image seg-mentation tests .The experimental results indicate that it not only improves the parallel computation of the image and reduces the complexity of the space and time ,but also provides new thoughts on applying the knowledge granulation to the image process .