摘要
针对传统的图像分割算法难以分割出具有语义特征的目标区域的问题,将形状描述子的识别能力引入图像分割过程,利用其提供的较高层的语义信息,生成边界模板来指导图像分割,同时使用遗传算法来处理边界模板的匹配过程,从而智能地选择和分割出目标区域。最后,在形状描述子对全局形状信息的约束下,又根据目标图像的局部特征进行分割边界的优化。试验结果证明,本方法成功地将形状描述子的描述和识别能力引入图像分割过程,实现了对目标图像的选择性分割,达到满意的分割效果。
Aiming at the difficulty of segmenting objects with semantic feature in traditional image segmentation algorithm, we introduce the identifying capability of shape descriptor into the process of image segmentation, with whose high level semantic information, to generate boundary template to direct image segmenting. At the same time, genetic algorithm is applied to deal with the matching process of the boundary template so as to intelligently select and segment the object region. With the constraint of shape descriptor to the global shape information, segmenting boundary can get optimized according to the local feature of object image. Testing results indicate that the introduction of shape descriptor and identify capability successfully realized the selective segmentation of object image and achieve satisfactory effects.
出处
《电子测量技术》
2007年第12期1-2,6,共3页
Electronic Measurement Technology
关键词
图像分割
形状描述子
遗传算法
image segmentation
shape descriptor
genetic algorithm
