A Knowledge-Based Search Strategy for Optimally Structuring the Terrain Dependent Rational Function Models
AbstractIdentifying the optimal structure of terrain dependent Rational Function Models (RFMs) not only decreases the number of Ground Control Points (GCPs) required, but also increases the accuracy of the model, by reducing the multi-collinearity of Rational Polynomials Coefficients (RPCs) and avoiding the ill-posed problem. Global optimization algorithms such as Genetic Algorithm (GA), evaluate the different combinations of parameters effectively. Therefore, they have a high ability to detect the optimal structure of RFMs. However, one drawback of these algorithms is their high computation cost. This article proposes a knowledge-based search strategy to overcome this deficiency. The backbone of the proposed method relies on the technical knowledge about the geometric condition of image at the time of acquisition, as well as the effect of external factors such as terrain relief on the image. This method was evaluated on four different datasets, including a SPOT-1A, a SPOT-1B, an IKONOS-Geo image, and a GeoEye-Geo imagery, using various number of GCPs. Experimental results demonstrate the efficiency of the proposed method to achieve a sub-pixel accuracy using few GCPs (only 4–6 points) in all datasets. The results also indicate that the proposed method improves the computation speed by 140 times when comparing with GA. View Full-Text
Share & Cite This Article
Jannati, M.; Valadan Zoej, M.J.; Mokhtarzade, M. A Knowledge-Based Search Strategy for Optimally Structuring the Terrain Dependent Rational Function Models. Remote Sens. 2017, 9, 345.
Jannati M, Valadan Zoej MJ, Mokhtarzade M. A Knowledge-Based Search Strategy for Optimally Structuring the Terrain Dependent Rational Function Models. Remote Sensing. 2017; 9(4):345.Chicago/Turabian Style
Jannati, Mojtaba; Valadan Zoej, Mohammad J.; Mokhtarzade, Mehdi. 2017. "A Knowledge-Based Search Strategy for Optimally Structuring the Terrain Dependent Rational Function Models." Remote Sens. 9, no. 4: 345.