Automatic radiation uniformity correction of multispectral imagery acquired with unmanned aerial vehicle

作者：Yang Guijun; Wan Peng; Yu Haiyang; Xu Bo; Feng Haikuan

来源：Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(9): 147-153.

DOI：10.11975/j.issn.1002-6819.2015.09.023

• Automatic
• Chromatic aberration
• Correlative functions
• Hue saturation values
• Multi-spectral imagery
• Pmultispectral imagery
• Remote sensing applications
• Scale invariant feature transforms

Unmanned aerial vehicles (UAVs) are an exciting new remote sensing tool capable of acquiring high resolution spatial data. In a data acquisition task, hundreds of images are acquired in a sortie due to the very small footprint of each image. The gray-scale value of homonymous points in adjacent-image pairs which is gathered by unmanned aerial vehicle in a sortie should be consistent on an -idealized condition. Actually, the gray-scale value of homonymous points always yields different results due to changes in illuminance, different shooting angle etc. In this study, we described a methodology for radiation uniformity correction among different images. The first step of the methodology was matching the images by scale invariant feature transform (SIFT) feature detector to extract homonymous points from which its gray-scale value in different images could be used to establish correlative function in the next step. The uncorrected image then was corrected by the correlative function. We also proposed a radiation uniformity correction workflow of adjacent-image pairs and image sequences. In the area of photogrammetry, the color consistency in visual form was more emphasized and many dodging method was proposed. But those methods can not be utilized in remote sensing application which emphasizes the radiation uniformity. In this paper, the color consistency adjusting based on histogram matching, the radiation uniformity correction in the original color space and the luminance consistency correction in the hue-saturation-value (HSV) color space in allusion to three-channel images was performed and the results was compared among those methods. The experimental results indicated that the color consistency correction based on histogram matching method can obtain a very good effect visually, but can also lead to severe grayscale gap. The radiation uniformity correction in the original color space was capable of achieving radiation consistency,but had little chromatic aberration in some case. The luminance consistency adjusting in the HSV color space was able to obtain a great result both visually and in radiation consistency, but the method was limited in utilization of three-channel images. The noise was enhanced after correction in some case. The bilateral filter was used to remove the noise, concurrently to retain radiation consistency between corrected image and reference image. The last part of this paper described the limitation of our method and indicated the needs for further studies. Also the processing chain of image sequences was proposed. Propagation of errors may cause more aberration at the image which far from reference image. So, the block adjustment may be considered into next step studies. In addition to this, SIFT algorithm was a time consuming method for this single target. So, we can integrate radiation uniformity correction into aero triangulation, both of which need feature extraction. Finally, the effect of sensor calibration on radiation uniformity correction should be studied in the future.

• 出版日期2015
• 单位Polytechnic university