Doorphate image quantization based on peer group filtering for a mobile robot
GS Yang, Min Han Tan, ZG Hou · 2004
Doorplate image processing for a mobile robot is studied in this paper. Firstly, considered a special case that all pixels have the same values in an image window, the basic Peer group filter (PGF) method is perfected. For each window, the average activity level over all pixels in the whole image is applied to estimate the threshold for determining the size of the peer group, which leads to a Simplified PGF (SPGF) algorithm to reduce the computational load of PGF algorithm. Secondly, according to the norm distance between the value of a pixel in the image processed by SPGF and the value of a given color frequently appearing in the background of the doorplate, all pixels in the whole image are coarsely divided into two groups based on the average activity level. In the first group, SPGF is performed on all pixels to form a cluster representing the background of the doorplate. The pixels that do not belong to this cluster are removed from the first group to the second group. The same method is performed on the second group to obtain another cluster representing the foreground of the doorplate. The remains of the image form the third cluster. In the last part, simulations are done to test the effectiveness of the presented algorithm.