基于改进分水岭算法的熏烤肉图像分割  

Smoked barbecue image segmentation based on improved watershed algorithm

作　　者：何鹏[1] 王成琳[1] 王福刚[1] 

机构地区：[1]齐齐哈尔大学通信与电子工程学院,黑龙江齐齐哈尔161006

出　　处：《电子技术应用》2012年第10期126-128,共3页Application of Electronic Technique

基　　金：黑龙江省普通高等学校青年学术骨干支持计划项目(1155G63);齐齐哈尔大学研究生创新项目(YJSCX2011-038X)

摘　　要：针对传统分水岭算法容易产生过度分割的问题,提出一种改进分水岭算法,并用来分割熏烤肉表面颜色。算法先对熏烤肉原始图像作滤波预处理,然后作传统分水岭变换,对产生的过度分割区域,在RGB颜色空间中进行自动种子选取及种子区域生长,最后对剩余小区域进行合并得到分割图像。实验表明,该方法减少了过度分割现象,成功地分割熏烤肉表面颜色,为之后的分析工作奠定了基础。To over-segmentation phenomenon of traditional watershed algorithm, an effective improved watershed algorithm is proposed in this paper to segment surface color of smoked barbecue. The algorithm firstly preprocess the original image by using fil- ter, then in RGB color space, to the over-segmentation regions of traditional watershed algorithm, the algorithm applies seed region growing based on automatic seed selection and region merging to reduce over-segmentation phenomenon and then gets the segmenta- tion image. Experiment shows that this approach reduces over-segmentation and segments surface color of smoked barbecue success- fully and lays a foundation for the analysis later.

关 键 词：计算机视觉 熏烤肉 分水岭算法 图像分割 种子生长 

分 类 号：TP317[自动化与计算机技术—计算机软件与理论]