| Autor: |
Zhu, Kexin, Ma, Xiaodan, Guan, Haiou, Feng, Jiarui, Zhang, Zhichao, Yu, Song |
| Předmět: |
|
| Zdroj: |
International Journal of Remote Sensing; Apr2021, Vol. 42 Issue 7, p2463-2484, 22p |
| Abstrakt: |
Abstract: Light distribution in the soybean canopy is affected by the leafstalk angle, which is of great significance for breeding high-quality soybean cultivars. In order to calculate the soybean leafstalk angle more accurately, three soybean varieties (Kangxian-9, Kangxian-13 and Fudou-6) were taken as studies objects, first, the colour and depth images of soybean canopy during reproductive growth period were obtained by Kinect V2.0 sensor. Second, the image of soybean canopy was recognized from original two dimensional image using k-means clustering segmentation algorithm. Third, the soybean canopy skeleton was extracted using morphology thinning method. Further, the branch points and endpoints of skeleton were detected in three dimensional space by optimized corner detection algorithm based on the registration between depth images and the corresponding colour images. Finally, the calculation of the leafstalk angle was realized according to the principle of angle calculation. The results showed that compared with the measured values, the coefficient of determination (R2) between calculated values and measured values for the three cultivars were 0.94, 0.92 and 0.84, respectively, which could meet the need of accuracy for breeding. Thus, the proposed method can not only calculate the leafstalk angle of soybean accurately, but also provide technical support for calculation of other phenotypic traits. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
