基于图像处理技术的5种红树林叶片形态特征及叶绿素相对含量的估测

为研究基于计算机视觉的图像处理技术获得的红树林叶片形态特征和叶绿素相对含量,以5种红树林为研究对象,分析其叶片形态特征及建立叶片颜色参数和叶绿素相对含量的回归模型,为苗木培育及营养诊断提供理论依据。结果表明：基于图像处理技术获得红树林叶片形态特征值是可信的,5种红树林叶片形态特征（叶面积、叶长、叶宽、叶周长、叶片形态因子）存在显著差异。红海榄和木榄单片叶片最大,其次是秋茄和桐花树,白骨壤叶片最小,白骨壤叶片形态接近椭圆形,红海榄和秋茄2种红树叶片属于长条型,木榄和桐花树叶片形态类似,红海榄叶长和叶宽相关性最强。5种不同种类红树林的SPAD值和Dualex值差异性显著。利用数码相机获取5种红树林叶片彩色图像,通过图像处理软件提取RGB以及通过差值、比值、标准化值、归一化值等运算组合的28种颜色特征值,与叶绿素相对含量（SPAD值、Dualex 值）进行回归分析,B值与SPAD值极显著相关,(R+B+G)/3、R/B、G/B与SPAD值显著相关;18种颜色特征参数与Dualex值显著相关,其中相关系数最大的为G-B和G值。因此,可依据颜色特征参数（R、G、B）建立相应的统计模型进行红树林叶片叶绿素含量的估测,进而为检测红树林生长、诊断营养状况以及进行水肥精确管理提供理论依据和技术支持。

Morphological Characteristics of Leaves and Estimation of Relative Chlorophyll Content in Five Species of Mangrove Based on Image Processing Technology

The morphological characteristics of mangrove leaves, the relationship between color characteristic parameters of leaves and chlorophyll content were acquired by the computer vision image processing technology. The leaf morphological characteristics and regression models of leaf color parameters and chlorophyll content were established to provide a theoretical basis for seedling cultivation and nutritional diagnosis. It was reliable to obtain morphological characteristic values of mangrove leaves based on the image processing technology. There were significant differences in leaf morphological characteristics (leaf area, leaf length, leaf width, leaf circumference and leaf morphological factors) among the five mangrove species studied. Rhizophora stylosa and Bruguiera gymnorrhiza had the largest single leaf, followed by Kandeliaobvolata and Aegiceras corniculatum, and Avicennia marina had the smallest leaf. The leaf morphology of A. marina was nearly elliptic. R. stylosa and K. obvolata belonged to long strip type. The leaf morphology of B. gymnorrhiza and A. corniculatum was similar, and the correlation between leaf length and leaf width of R. stylosa was the biggest. The SPAD value and Dualex value of the five different mangrove species were significantly different. Five kinds of color images of mangrove leaves were acquired by a digital camera. RGB was extracted by an image processing software and 28 kinds of color eigenvalues were combined by difference, ratio, standardization and normalization, and the relative content chlorophyll (SPAD value, Dualex 4 value) was analyzed by regression. Statistical test showed that there was extremely significant correlation between B value and SPAD data, (R+B+G)/3, R/B and G/B were significantly correlated with SPAD data, and 18 color characteristic parameters were significantly correlated with Dualex data, among which G-B and G values had the largest correlation coefficient. Therefore, statistical model could be established to estimate the chlorophyll content of mangrove leaves based on the color characteristic parameters, which could provide a theoretical basis and technical support for the detection of mangrove growth, diagnosis of nutritional status and precise management of water and fertilizer.