Study On Kidney Beans Yield Estimation Based On UAV Remote Sensing Image

China is the main exporter of kidney bean in the world because of its rich germplasm resources and long history of cultivation.Because the phenotypic traits of kidney bean vary greatly in different growth stages,manual measurement is time-consuming and laborious.And there are human errors,and the acquisition of phenotypic traits is the basis of breeding work,so it is very important to monitor the growth of crops in real time,effective field inspection and management and accurate yield estimation.With the development of precision agriculture,remote sensing technology has a wide range of coverage,rapid update speed,flexibility,and high-throughput access to agricultural information,which has laid a foundation for rapid and efficient crop monitoring and yield estimation.Based on the remote sensing image data of 15 kidney bean varieties planted in Heilongjiang Province at flowering,flowering and pod stages and filling stages,the vegetation index extracted was studied.The relationship between chlorophyll content and kidney bean yield was studied.Based on the optimal growth period of yield prediction,linear regression and partial least square regression(PLSR)were used to model the kidney beans yield.The determination coefficient(R~2),root mean square error(RMSE)and normalized root mean squared error(NRMSE)were used as testing indexes to analyze the precision of kidney bean yield estimation model based on single vegetation index and multiple vegetation index.The main findings were as follows:(1)The chlorophyll content of kidney beans showed a trend of"low-high-low"from the flowering stage to the grain-filling stage.The five vegetation indices NDVI,RVI,GNDVI,EVI and NDRE were correlated with chlorophyll content.Chlorophyll contents were all correlated.The results showed that NDVI and NDRE had extremely significant correlations with chlorophyll contents in the three growth stages,and the highest correlation coefficient between NDVI and chlorophyll content was 0.47 at the flowering and pod stage.(2)The correlation analysis of kidney bean yield of five vegetation indices was conducted.The results showed that the correlation between NDVI and yield was the highest at flowering and pod stage.The R~2of kidney bean yield estimation model based on NDVI was 0.55,RMSE was253.41 kg/hm~2and NRM SE was 19.49%.At the same time,the effect of the models based on the single vegetation index is better than the flowering and filling stages.The best growth period is the flower and pod stage.(3)Through the correlation analysis of vegetation indices,the highest correlation between NDVI and NDRE is 0.88.The yield estimation model of kidney bean based on NDVI and NDRE combination was validated.The results showed that the yield prediction model based on multi-vegetation index combination was the most accurate in flowering and pod stage.The determination coefficient R~2was 0.81,the RMSE was 223.78 kg/hm~2and the NRM was 17.21%.The effect of the model was better than that of the kidney bean yield estimation model constructed by single vegetation index.