Hiroshi SAITO
A Study on Canopy Green Measurement for Paddy Rice Using Digital Cameras Loaded onto a Small Unmanned Aerial Vehicle
Kazuyoshi TAKHASHI, Atsushi RIKIMARU, Kenta SAKATA
One of the quantities of growth of the paddy rice includes a leaf color. Confirms the nitrogen nutritional status of paddy rice from the leaf color, prevent water pollution by performing proper fertilization, it is possible to produce high quality rice. A leaf color scale and a chlorophyll meter for carrying are used for the leaf color measurement. Leaf color scale corresponds to canopy green measurement and monoplane measurement. However, objectivity is low to judge it from viewing. The chlorophyll meter can grasp a nourishment state of the paddy rice objectively, and plasticity is high. However, it can only be done monoplane measurement. Therefore, seeking a highly reproducible extensive measurement method, method using remote sensing has been studied. In addition, UAV (Unmanned Aerial Vehicle) is beginning to be used in a variety of applications. If it is canopy green measurement for paddy rice using a UAV, thereby enabling a wider ranges of measurement. However, a pixel value changes by an illumination condition and setting at the time of the photography even if an object is the same.
In the present study, it was at the same time photographed the color is uniform criteria plate there is no property change with time and paddy rice. Captured image is a reproduction process, converted a pixel level and the spectral reflectance of the criteria plate into reflectance in a linear equation, a quadratic equation, straight line interpolation for the cause. As a result, it was determined that the method using the linear interpolation is most suitable. Then, to resemble the pseudo-paddy rice the leaf color scale, it was studied reproducibility of leaf color. The difference between the spectral reflectance of leaf color scale was 6-7% in the red band, 6-9% in the green band, about 2% in blue band. The standard deviation of the reflectance of the time was 3-5%. When excluding the date a large variation in reflectivity, the difference between the spectral reflectance of leaf color scale was 5-7% in the red band, 5-8% in the green band, about 1% in blue band. The standard deviation of the reflectance of the time was 1-2%. In addition, the reflectance of each band of leaf color scale as coordinates in RGB color space was confirmed using linear discriminant obtained from discriminant analysis whether the identification can be numbers. As a result, No.3 and No.4 is there is a date that is not able to identify, No.4 and No.5 were able to identify at all dates. It is considered the difference between the spectral reflectance of each number is affected.
Please use "previous botton" to return the previous page