摘要： 针对传统种子区域生长算法在分割具有复杂纹理的彩色图像中存在的问题，提出一种改进的种子区域生长算法，该算法在YCbCr颜色空间中进行，采用离散余弦变换提取图像纹理特征值，进行自动种子及种子区域的生长，并用区域合并改善过度分割。实验结果表明，该算法能有效提高图像分割的精确性。

关键词: 种子生长, 区域生长, 区域合并, 彩色图像分割

Abstract: Color image with complicated texture is insufficiently dealt with a traditional seed region growth method. Aiming at this problem, an improved seeded region growth method to segment color image is presented. The research is based on YCbCr model. The texture features are obtained by discrete cosine transform. The seeds and seeded region are produced automatically. The region-merging is used to improve the image segmentation result. Experimental results show the algorithm can effectively improve segmentation accuracy for different image natures.

Key words: seed growth, region growing, region merging, color image segmentation

中图分类号: 

• TP311.52