基于改进分水岭算法的熏烤肉图像分割

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。

Smoked barbecue image segmentation based on improved watershed algorithm

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 filter,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 segmentation image.Experiment shows that this approach reduces over-segmentation and segments surface color of smoked barbecue successfully and lays a foundation for the analysis later.