Proposing a New Method for Detecting Single and Cluster Blunders of Digital Elevation Model Based on the Genetic Algorithms

Digital Elevation Models (DEMs) are one of the most important kinds of geospatial data in Geospatial Information Systems (GIS). DEMs contain different types of errors caused by sampling, measurement, and reconstruction processes. So accuracy is considered as one of the important attributes of DEMs. The errors can be classified into three types including random, systematic, and gross. This paper presents a new algorithm for detecting gross errors of irregular DEM data using Artificial Intelligence (AI) algorithms. The proposed gross error detection method are characterized by a common localization procedure: entire dataset was examined by consideration iteratively only a small subset at a time; for each step the data belonging to a moving square window was taken into account and each dataset was separately handled. In each window a bilinear surface was fit to fully surrounded points and the residual for each point was estimated. By the use of Genetic algorithm, it was tried to minimize the sum of squared residuals in order to detect points with gross error. The results showed that the proposed gross errors detection method is more effective and feasible than existing algorithms such as data snooping.