摘要
探讨了利用图像识别技术快速获取棉花水分信息的方法。分析了颜色参数与棉花水分含量及水分含量指数的关系,并建立了水分状况的预测模型。结果表明,HIS颜色系统的色调H值与棉花含水量显著正相关,而亮度I值和水分含量指数呈极显著正相关;RGB颜色系统的G-R与棉花水分含量及水分含量指数相关性最好。以G-R参数和棉花水分含量及水分含量指数建立回归模型,预测精度分别达到了90.71%和91.02%,预测模型分别为0.3955+0.0139(G-R)(R2=0.7460**)和-0.0428+0.0223(G-R)(R2=0.8552**)。基于图像识别技术诊断棉花水分状况是可行的,有望成为作物水分信息获取的新手段。
This paper tries to find the method of obtaining the water information of cotton population by image recognition technology. The correlation between color parameters of cotton population digital images and water content and water content index were analyzed, and predicted models was established. The results showed there were significant correlations between Hue and Intensity in the HIS color system and water content and water content index at the 0.05 and 0.01 probability levels, respectively. G-R in the RGB system was the best parameter to predict water content and water content index and established regression models. Both of the accuracy of estimated models was about 90.71% and 91.02%. The predicted models was 0.3955 ＋ 0.0139（G - R）（R^2 = 0.7460 ^＊＊）and - 0.0428 ＋ 0.0223（G - R）（R^2 = 0.8552^＊＊） .It is feasible to diagnose the water status of cotton population by image recognition technology and it is hopeful to be a new means of obtaining crop water information.
出处
《石河子大学学报（自然科学版）》
CAS
2007年第4期404-407,共4页
Journal of Shihezi University(Natural Science)
基金
"863"项目资助(2006AA10A302)(
2006AA10Z207)
国家自然科学基金(30360047)
关键词
棉花
图像识别
含水量
水分含量指数
图像覆盖度
cotton
image recognition
water content
water content index
percent ground cover of image
