Automatic Conversion of DSM to DTM by Classification Techniques Using Multi-date Stereo Data from Cartosat-1

Abstract

Stereoscopic satellite imaging provides directly the digital surface model (DSM) in contrast to the topographic application which needs digital terrain model (DTM). Presently, DTM is generated manually in 3d environment which is a time consuming exercise. To address this problem, the paper proposes DTM generation from Cartosat-1 stereo by adopting automatic classification of mass points similar to LiDAR points by improving its density using multi-date Cartosat-1 stereo data. To illustrate the proposed methodology, two study areas were investigated viz. Dharamshala, Himachal pradesh and Netravathi river,Western ghats, Karnataka in India. The results from the proposed methodology demonstrated that classification techniques when applied to the densified mass points from multi-date stereo data, works well in comparison to mass points generated from a single scene. The study also demonstrated that terrain parameters like slope, drainage pattern and contour derived from the DTM generated by the proposed method were similar to DTM generated by manual methods.