基于二维直方图的彩色图像全局阈值分割方法研究

Research on Global Threshold Segmentation Method of Color Image Based on Two-Dimensional Histogram

当前彩色图像分割采用手动选择种子点随机性大,且未考虑轮廓生长规则存在分割准确度低的问题.故此,文中提出基于二维直方图的彩色图像全局阈值分割方法.建立彩色图像全局阈值分割的边缘轮廓特征提取模型,采用水平集法获取彩色图像的初始轮廓;结合二分法确定彩色图像全局轮廓生长规则,构建彩色图像规划的视觉图像采集模型;采用高分辨灰度直方图特征提取方法进行彩色图像的视觉信息增强处理,构建彩色图像的多维特征提取和关键特征点检测模型;采用小波多尺度特征分集方法实现彩色图像的分布式重组和二维直方图重构,建立彩色图像的全局阈值分割模型;通过统计形状分析法实现对彩色图像的优化分割.仿真结果表明,采用该方法进行彩色图像分割的自适应性较好,分割后的彩色图像种子点区域具有高度的连通性,提高了分割的精度,从而提高了彩色图像的视觉表达和特征提取能力.

At present,the color image segmentation adopts manual selection of seed points with large randomness,and the problem that the segmentation accuracy is low is not considered in the contour growth rule.Therefore,a global threshold segmentation method for color images based on two-dimensional histogram is proposed.The edge contour feature extraction model of global threshold segmentation of color image is established.The initial contour of color image is obtained by level set method.The global contour growth rule of color image is determined by combining dichotomy.The visual image acquisition model of color image planning is constructed,and high resolution gray scale is adopted.The histogram feature extraction method is used to enhance the visual information of color images,construct the multi-dimensional feature extraction and key feature point detection model of color images,and adopt the wavelet multi-scale feature diversity method to realize the distributed reorganization of color images and the reconstruction of two-dimensional histograms.The global threshold segmentation model of color images is established,and the optimal segmentation of color images is realized by statistical shape analysis.The simulation results show that the color image segmentation has better adaptability,and the segmented color image seed point region has high connectivity,which improves the segmentation accuracy and improves the visual expression and feature extraction ability of color images.