基于最小生成树的数字图像多层次分割方法研究

Research on Multi-level Segmentation Method of Digital Image Based on Minimum Spanning Tree

如今,由于图像离散化造成分割的误差大,因此,提出基于最小生成树的数字图像多层次分割方法的研究。首先,基于最小生成树进行数字图像预处理,对图像进行整体特征提取,其次,构建出图像多层次分割标准,以不同的场景需求为依据,来对权值进行定义,确保最终的分割结果在内部具有高度的相似性,最后,实现数字图像多层次分割。为验证文章方法的有效性需要进行对比测试,实验结果:基于最小生成树的数字图像多层次分割方法的区域数目较少更符合人类视觉特征,分割效果更加理想,具有较好的性能。

Nowadays,due to the large segmentation error caused by the image discretization,the rescarch of the multi-level scgmentation method of digital image based on the minimum genera-ting tree is proposed.First of all,based on the minimum generation tree for digital image pre-proccssing,the overall image of feature cextraction,secondly,build the image multi-level seg-mentation standard,based on different scene requirements,to define the weights,to ensure that the final segmentation results within a high degree of similarity,finally,realize the digital image multi-level segmentation.In order to verify the effectiveness of the method in the paper,a comparative test is needed.The experimental results:the number of digital image based on the minimum spanning trce is more in line with human visual characteristics,the segmentation effect is more ideal,and has good performance.