Automatic feature detection based on classification of combination of orthoimage from digital camera and Lidar DSM

Feature detection methodologies are very important in the context of spatial data capture and updating for GIS applications. These methodologies can be based on either LIDAR or photogrammetric data or even on a combination between them. Firstly true orthoimages were generated by orthorectification of digital camera images using DSM from LIDAR data .Leica Photogrammetric Suite (LPS) module of Erdsa Imagine 9.2 software was utilized for processing. Secondly supervised maximum likelihood classification was performed using three different feature sets. The first classification was performed using the three digital aerial image channels while the second classification was performed using the three digital aerial image channels and two LIDAR feature images(average calculated from first pulse height data and standard deviation calculated from first pulse height data ).The third classification was performed using two LIDAR feature images. The overall accuracy of the first approach was 91%, and kappa coefficient was 0.82, the overall accuracy of the second approach was 93%, and kappa coefficient was 0.85 and the overall accuracy of the third approach was 79%, and kappa coefficient was 0.76. After that morphological operations were performed in order to remove noise. It was found that the second approach is better than the first followed by the third approach.