As a kind of new survey equipment with high detection accuracy and rich data information，laser radar is gradually applied to the information monitoring field about inland river shoreline.Because the length of shoreline is greater than the measuring range of the laser radar，it is necessary to continue scanning different locations and get the complete shoreline outline.To solve the registration problem of multi-location point cloud data caused by continuous scanning，this paper implements the ICP (Iterative Closet Point) registration algorithm based on the KD-Tree data structure.The proposed algorithm not only concentrates multi-location of point cloud data on a unified coordinate system，but also reduces effectively the data search scope of ICP registration algorithm and improves the accuracy of registration.The results of field tests show that the algorithm can reduce effectively the registration time and achieve corresponding registration accuracy.On the whole，the algorithm implements good processing results of point cloud registration.