A GROUND REFERENCE TARGET DETECTION METHOD FOR AUTOMATIC VICARIOUS CALIBRATION OF UAV MULTISPECTRAL IMAGE

摘要

Recently, demands of unmanned aerial vehicle (UAV) image has increased in various fields, and sensor technology are rapidly developing. Radiometric correction is needed to transform from pixel's digital number (DN) to spectral reflectance when using spectral information in an application. To obtain spectral reflectance, some sensors for UAVs measure a reference target before taking-off, and measure the irradiance during image acquisition. For sensors without the irradiance, spectral reflectance can be obtained by vicarious radiometric calibration with scenes including ground reference target. Currently, for vicarious calibration, the commercial software is designed to manually specify the area of the ground reference target in the scene. In this study, we propose an automatic method to detect a ground reference target in a scene for vicarious calibration of multispectral UAV image without radiometric calibration parameters. All of reference targets were detected in the experiment condition. Further study is needed to decrease false alarm with similar spectral and morphological characteristics.