|WU, MINGQUAN - Institute Of Remote Sensing And Digital Earth, Chinese Academy Of Sciences
|SONG, XIAOYU - Institute Of Remote Sensing And Digital Earth, Chinese Academy Of Sciences
|HUANG, WENJIANG - Institute Of Remote Sensing And Digital Earth, Chinese Academy Of Sciences
|NIU, ZHENG - Institute Of Remote Sensing And Digital Earth, Chinese Academy Of Sciences
|WANG, CHANGYAO - Institute Of Remote Sensing And Digital Earth, Chinese Academy Of Sciences
|WANG, LI - Institute Of Remote Sensing And Digital Earth, Chinese Academy Of Sciences
Submitted to: Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/2/2017
Publication Date: 3/4/2017
Citation: Wu, M., Yang, C., Song, X., Hoffmann, W.C., Huang, W., Niu, Z., Wang, C., Wang, L. 2017. Evaluation of orthomosics and digital surface models derived from aerial imagery for crop mapping. Remote Sensing. doi:10.3390/rs9030239.
Interpretive Summary: Aerial imagery acquired by consumer-grade cameras is being increasingly used for crop mapping and assessment. This study used imagery from a normal color camera and a near-infrared camera to measure crop height and identify crops. Crop height was extracted from the digital surface models generated from aerial images. Two image classification methods were used to identify crop types from the mosaicked imagery and crop height. Results showed crops could be identified with overall accuracy values of 78.5% and 97.5% depending on the classification method. The findings from this study indicate that classification methods have a more significant effect on classification results than canopy height, though additional height information from aerial imagery has the potential to improve crop classification accuracy.
Technical Abstract: Orthomosics derived from aerial imagery acquired by consumer-grade cameras have been used for crop mapping. However, digital surface models (DSM) derived from aerial imagery have not been evaluated for this application. In this study, a novel method was proposed to extract crop height from DSM and to evaluate the orthomosics and crop height for crop mapping. Crop height was extracted by subtracting the DSM derived during the crop growing season from that derived after crops were harvested. Then crops were identified from four-band aerial imagery (blue, green, red and near-infrared) and crop height using an object-based classification method and a maximum likelihood method. The results showed that the extracted crop height had a very high linear correlation with field measured crop height with an R-squared value of 0.981. For the object-based method, crops could be identified from the four-band airborne imagery and crop height with an overall accuracy of 97.50% and a kappa coefficient of 0.9545, which were 2.52% and 0.044 higher than those without crop height. For maximum likelihood, crops could be mapped from the four-band airborne imagery and crop height with an overall accuracy of 78.52% and a kappa coefficient of 0.6658, which were 2.63% and 0.034 higher than those without crop height.