茶叶的生长情况和成熟度是影响茶叶品质的重要因素,而完成茶叶分析和识别的基础是对茶叶图像的准确分割。文章提出一种基于茶叶颜色和种子区域生长的改进方法来完成对茶叶嫩芽的分割。首先将原始RGB彩色图像转化为HSI空间,选取H和S参数进行初步的种子选择,然后对种子区域基于颜色的相似性和区域的邻接性进行区域生长,并结合颜色距离和边缘距离进行区域生长和合并。文中对不同角度取像图像的茶叶嫩芽进行了分割和比较,试验结果表明该算法能很好地将茶叶嫩芽从茶叶中分离出来,并较好地保存了茶叶嫩芽的轮廓信息。
Growth conditions and degree of maturity are two important factors affecting the quality of tea, while accurate segmentation of tea images is the precondition to analyze and identify the tea leaves. In this paper, an improved method based on the color and regional growth of tea is introduced to conduct the image segmentation of tender tea shoots. Firstly, the RGB color images are converted into HSI space, and then H and S parameters are selected for the initial selection of tea seed. Secondly, the regional growth based on color similarity and regional adjacency are conducted on the seed region, and then the color and edge distance is used to carry out regional growing and mergence. The image segmentation of tender tea shoots acquired from the different perspectives are conduced and compared, and the results indicate that this algorithm performs well in separating tender shoots from leaves and maintaining the outline of tender shoots.
[1] 陆建良, 梁月荣, 龚淑英, 等. 茶汤色差与茶叶感官品质相关性研究[J]. 茶叶科学, 2002, 22(1): 57-61.
[2] 蔡健荣. 利用计算机视觉定量描述茶叶色泽[J]. 农业机械学报, 2000(4): 67-70.
[3] 汪建, 杜世平. 基于颜色和形状的茶叶计算机识别研究[J]. 茶叶科学, 2008, 28(6): 420-424.
[4] 林开颜, 吴军辉, 徐立鸿. 彩色图像分割方法综述[J].中国图像图形学报, 2005, 10(1): 1-10.
[5] GONZALEZR C.数字图像处理[M]. 阮秋琦, 译. 北京: 电子工业出版社, 2003.
[6] Fan J, Yaud K Y, Ekmagarmid A K, et al. Automatic image segmentation by integrating color-edge extraction and seeded region growing[J]. IEEE Transactions on Image Processing, 2001, 10(10): 1454-1466.
[7] Mehnert A, Jackway P.An improved seeded region growing algorithm[J]. Pattern Recognition Letters, 1997(18): 1065-1071.
[8] Comaniciu D, Meer P.An algorithm for data-driven bandwidth Selection[J]. IEEE Trans PAMI, 2003, 24(5): 281-288.
[9] 叶齐祥, 高文, 王伟强, 等. 一种融合颜色和空间信息的彩色图像分割算法[J]. 软件学报, 2004, 15(4): 522-527.