The purpose of image-map registration is to revise the digital map included in Geographic Information System (GIS) with Remote Sensing Images (RSI). Because edge points extracted from RSI are very different from those extracted from digital maps, the error of registration different edge points propagates to the final registration results. In this paper, we provide an automatic algorithm for the image-map registration problem. A novel concept of modified partial Hausdorff distance (MPHD) is proposed as a distance measure. This measure is quite tolerant of the differences between map and image. In order to decrease the computational complexity, we also use Hilbert curve to select the initial candidates of edge points in our algorithm. The number of initial candidates can be adjusted to meet the required search accuracy and/or speed. The experimental results implicate that MPHD of our algorithm could provide sufficient information for image-map registration.