Radiometric Normalization of Large Airborne Image Data Sets Acquired by Different Sensor Types

Abstract

ABSTRACT:Generating seamless mosaics of aerial images is a particularly challenging task when the mosaic comprises a large number of images, collected over longer periods of time and with different sensors under varying imaging conditions. Such large mosaics typically consist of very heterogeneous image data, both spatially (different terrain types and atmosphere) and temporally (unstable atmospheric properties and even changes in land coverage). We present a new radiometric normalization or, respectively, radiometric aerial triangulation approach that takes advantage of our knowledge about each sensor's properties. The current implementation supports medium and large format airborne imaging sensors of the Leica Geosystems family, namely the ADS line-scanner as well as DMC and RCD frame sensors. A hierarchical modellingwith parameters for the overall mosaic, the sensor type, different flight sessions, strips and individual images -allows for adaptation to each sensor's geometric and radiometric properties. Additional parameters at different hierarchy levels can compensate radiometric differences of various origins to compensate for shortcomings of the preceding radiometric sensor calibration as well as BRDF and atmospheric corrections. The final, relative normalization is based on radiometric tie points in overlapping images, absolute radiometric control points and image statistics. It is computed in a global least squares adjustment for the entire mosaic by altering each image's histogram using a location-dependent mathematical model. This model involves contrast and brightness corrections at radiometric fix points with bilinear interpolation for corrections in-between. The distribution of the radiometry fixes is adaptive to each image and generally increases with image size, hence enabling optimal local adaptation even for very long image strips as typically captured by a line-scanner sensor. The normalization approach is implemented in HxMap software. It has been successfully applied to large sets of heterogeneous imagery, including the adjustment of original sensor images prior to quality control and further processing as well as radiometric adjustment for ortho-image mosaic generation.

Radiometric Correction of Close-Range Spectral Image Blocks Captured Using an Unmanned Aerial Vehicle with a Radiometric Block Adjustment

“…The outputs of the process are the radiometric model parameters, which are used to produce radiometrically corrected image products, such as reflectance mosaics, reflectance point clouds or multi-view and -angular reflectance observations of objects of interest. Similar approaches have previously been used with aircraft images [39][40][41][42]. We have used the radiometric block adjustment method already in various applications, including for agricultural and forest use [12,[32][33][34][35][36].…”

From research to production: radiometric block adjustment to automate and improve tonal adjustment of orthomosaics created from airborne images

“…Various aspects affect to the colour, or tones of the images collected from manned and unmanned aircrafts and form satellites: time of the day and year, atmosphere, illumination conditions, view and illumination angle, BRDF-effects (Bidirectional reflectance distribution function), object, sensor and whole imaging system (Chandelier andMartinoty, 2009, Li et al 2019). Due to these reasons, creating an evenly coloured mosaic from multiple, overlapping images is a challenge (Gehrke and Beshah, 2016).…”

On-Orbit Radiometric Calibration for a Space-Borne Multi-Camera Mosaic Imaging Sensor

“…The processing results showed that the overall accuracies were improved by up to 16.5% in the results of intensity data classification [25]. For generating seamless mosaics of aerial images obtained by different sensors, Gehrke and Beshah used the RBA to compensate radiometric differences based on RTPs, RCPs and image statistics [26]. Based on atmospheric radiative transfer (ART) models, pre-selected BRDF models and RCPs, the RBA was implemented for the digital radiometric model (DRM) and orthophoto mosaics showing no radiometric differences at the seam lines [27].…”