关键词: OPENCV, 青冈栎, 叶绿素含量, 机器视觉, 多元回归分析

Abstract: In order to obtain plant chlorophyll content in real time, by convenient and economic way, this paper studied the real-time detection system for cyclobalanopsis glauca chlorophyll content based on machine vision library OPENCV. Firstly, blade image was acquired through a digital camera, image R, G, B value was obtained after dealing with threshold segmentation, noise processing and image traversal. Then, different image color characteristic parameters were got from a variety changes combination of image R, G, B. The correlation of image color feature parameters and chlorophyll content of cyclobalanopsis glauca were analyzed, and high correlation coefficient of leaf image color characteristic parameters and chlorophyll content were analyzed through fitting analysis. The results showed that the image feature parameters R, R-B, (R-B)/(R+B) were very significantly correlated. Based on that, the chlorophyll content detection model was established. In addition, the detecting system was written by C++, OPENCV and QT4. Finally, the system detecting results were compared with that by the other methods; average error of system detecting result was found out to be 7.19%; and maximum error was 12.65%, proving the validity and accuracy of this system.

Key words: OPENCV, Cyclobalanopsis glauca, chlorophyll content, machine vision, multiple regression analysis