A robust and fast Hausdorff distance (HD) method is presented for image matching. Canny edge operator is used for extracting edge points. HD measure is one of efficient measures for comparing two edge images by calculating the interpixel distance between two sets of edge points, and does not require the point-to-point correspondence. However, high computational complexity is a common problem for HD measure because a large number of edge points could be extracted used to calculate HD. Further, a great many incorrect edge points will be extracted under the condition of occlusion and other ill conditions. A gradient orientation selectivity strategy is proposed to not only select available edges, but also reduce the number of edge points. Experimental results show that the proposed method has less computational cost, and has good robustness for object matching, especially under partial occlusion and other ill conditions.