Brickpath region detection using color and shape pattern information

This paper proposes a method for detecting brick path region in the outdoor pavement environment based on the analysis of color and shape pattern. The pavement typically consists of bricks with specific shape and colors. Furthermore, bricks form repeated structure. Thus the properties play an important role in recognition process. First, the image is segmented by color in order to detect candidate regions with similar color properties. Then the shape pattern detection based on edge structure of the block boundary is applied. The intersection point is extracted through the edge. The erroneous extracted intersection points are filtered out using angle histogram between intersection point and the center of gravity in each block. Finally each block has a perimeter vector as shape pattern and the similar blocks are detected as brick path. The experimental results are described in section 3 and they are analyzed that true positive is 86.52%.