In this paper, a novel method is proposed to extract rocks from Martian
surface images by using 8 data field. It models the interaction between two
pixels of an image in the context of imagery 9 characteristics. First,
foreground rocks are differed from background information by binarizing 10
image on roughly partitioned images. Second, foreground rocks are grouped into
clusters by 11 locating the centers and edges of clusters in data field via
hierarchical grids. Third, the target 12 rocks are discovered for the Mars
Exploration Rover (MER) to keep healthy paths. The 13 experiment with images
taken by MER shows the proposed method is practical and potential.Comment: This paper has been withdrawn by the author due to a crucial sign
error in equation