Automatic feature extraction from aerial images
The objective of this study would be to explore the possibilities of automatic feature extraction through the fusion of high resolution aerial digital camera images and DSM either from Lidar technology or from photogrammetry. As per the literature, the fusion of DSM from Lidar scanners and image data from digital cameras showed promising results. The data from these two sources is complementary to each other and hence make a great combination
Segmentation of Aerial / satellite data
Segmentation becomes more important with increasing spatial resolution of imagery. Texture in high-resolution aerial and high resolution satellite images requires substantial amendment in the conventional segmentation algorithms. The potential applications of this segmentation process are (1) Automatic 3D model generation (2) automatic DEM generation from DSM (3) Automation in Quality Checking of vector maps and many more
Development and incorporation of lossless compression algorithm(s)/ technique(s)
Development and incorporation of lossless compression algorithm(s)/ technique(s) for better handling, archival, effective usage and time critical data processing of aerial analog and digital data. The compression algorithm must meet the requirement of aerial compression system for storage and operation time. The compression techniques must reduce operation memory, and accelerate the processing speed without any data loss.
Radiometric calibration
Radiometric calibration refers to the statistical process of deriving physically welldefined radiometric quantities from the sensor’s raw measurements. Waveform LiDAR system provides detail information about the backscattering properties of the observed targets by recording complete return signal. Even though waveform measurements have high content of information, data acquired during different flight campaigns cannot be directly compared without a proper radiometric calibration. Model waveform Radiometric calibration requires formulation of LiDAR equation. This model may be useful for the retrieval of higher level data products.
