基于多层卷积滤波与HSV颜色提取的茶轮斑病识别研究
2018-07-18陈钊王子辉赵玉清高彦玉时玲
陈钊 王子辉 赵玉清 高彦玉 时玲



摘要:利用卷积滤波对原始图像进行显示轮廓以及锐化处理,引入了HSV模型进行颜色提取,通过二值化再次滤波完成了茶轮斑病斑区的提取,实现茶叶的茶轮斑病的快速诊断。结果表明,采用卷积滤波与HSV模型等图像处理技术能够较好地识别茶轮斑病的病斑区,对于图像清晰、面积大的病斑区识别比较精确。利用Python语言的嵌入性,为进一步实现茶叶茶轮斑病的精准喷雾系统打下了基础。
关键词:卷积滤波;图像处理技术;图像滤波;茶叶病检测;识别
中图分类号:S126 文献标识码:A 文章编号:0439-8114(2018)11-0107-04
DOI:10.14088/j.cnki.issn0439-8114.2018.11.027
Abstract: The original image is depicted and sharpen by convolution filtering, and then put it in HSV model to obtain the image color extraction. In order to get the target area of the tea wheel spot,it is also used the image binaryzation to complete the refiltering.Finally,that is a picture rapid diagnosis of tea leaf spot disease. The results showed that the image recognition of tea leaf spot in leaves of tea leaves was finished. By using convolution filtering and HSV model image processing technology can be better identified the round spot of tea lesion area, especially the clear image,size of lesion area. Taking advantage of the embeddedness of Python language,it lays the foundation for the accurate spray system of tea tea wheel spot disease.
Key words: convolution filter;image processing technology;image filtering;tea disease detection;identification
茶樹在中国是一种重要的经济类作物,种植面积多达3 529.0万hm2,达到了世界茶园面积的50%以上,遍及18个省。在茶树生长过程中,各个部分均易受到病虫的为害[1,2]。据统计,已知的茶树病虫害有100多种,其中有30多种比较常见。病虫害种类多、较严重的发生在茶芽与茶叶部位,直接造成的茶叶产量损失大概为总产量的15%~20%,局部地区或者个别年份损失比例更大。对于病虫害的防治,茶农一般选择化学农药。不仅是因为与无公害和绿色农药相比,化学农药价格低廉,还因为化学农药操作简单、见效快、效果可靠。化学农药的使用对于茶叶的高产、优产具有重要促进作用,但同时也会带来环境污染、农药残留等不利因素[3]。农药的低效率与滥用也导致中国茶叶农药残留量超标[4]。在众多茶叶病虫害中,茶轮斑病是茶园中比较常见的叶部病害,时有发生。……
