摘要
图像分割是图像识别过程中的一个重要步骤,在计算机视觉研究中占有非常重要的地位,图像分割的好坏直接影响图像识别的效果。为提高大规模图像数据集的分割效果,实现自动、快速、高质量分割,首先采用均值漂移算法对大规模图像进行预分割以降低图像规模,然后根据图像的颜色特征使用层次聚类算法对分割后的小区域进行聚类处理,以实现快速分割图像。实验结果表明,该方法能够有效减少图像分割时的运算时间和空间复杂度,提高图像分割效率,获取良好的图像分割效果,为后续图像分析、理解和识别打下基础。
Image segmentation, as an key part in the process of image recognition, performs the important effect on the research of computer vision because the image segmentation will affect the result of image recognition. To improve the segmentation effect of large scale image datasets and realize segmentation in automatic, high speed and high quality, this paper first uses the mean shift algorithm for large scale image datasets segmentation to reduce the image size, then, according to the color of the images features, uses hierarchical clustering algorithm to perform clustering processing for the small area after segmentation in order to quickly achieve segmentation of images. The test results show that this method can effectively reduce the operation time and space complexity, improve the efficiency of image segmentation, obtain good image segmentation effect, and lay the foundation for analysis, understanding and recognition of the image.
出处
《微处理机》
2015年第4期61-63,68,共4页
Microprocessors
基金
河南省科技厅科技发展计划项目(134300510037)
关键词
均值漂移算法
层次聚类算法
大规模图像数据集
图像平滑
预分割
图像识别
Mean Shift Algorithm
Hierarchical Clustering Algorithm
Large Scale Image Datasets
Image Smooth
Pre - segmentation
Image Recognition
