Crater Detection Using the Morphological Characteristics of Chang'E-1 Digital Elevation Models

This letter presents a new method for crater detection based on infrequently used morphological characteristics. Pixels representing the center and rim of each crater are connected and used to deduce its location and size. Based on the different characteristics shown by simple bowl-shaped and complex craters, two strategies are developed. The centers and rims of simple craters are detected directly using the original digital elevation models (DEMs). When it comes to complex ones, the slopes of DEMs and their derivatives are extracted to replace the original models. Methods for removing false craters are also developed. The characteristics of the new crater detection algorithm are as follows: 1) It avoids some deficiencies of previous methods and thus can detect new craters to complete current crater catalogs; and 2) the efficiency of the method is guaranteed due to its low time complexity. The new algorithm performs better when compared with the flooded object-oriented approach using Chang'E-1 DEMs in the same sample region, and it extends the LU60645GT catalog by 26.8% and 37.5% in two distinct areas when the crater diameters are larger than 6 km.