In this paper, we present scalable data parallel algorithms for geometric hashing. We perform implementations of the proposed algorithms on MasPar MP-1/MP-2. In earlier parallel implementations, the number of processors employed is independent of the size of the scene but depends on the size of the model database which is usually very large. We design new parallel algorithms and map them onto MP-1/MP-2. These techniques significantly improve upon the number of processors employed while achieving superior time performance. Our implementations run on a P processor machine, such that 1 ≤ P ≤ S, where S is the number of feature points in the scene. Our results show that a probe of the recognition phase for a scene consisting of 1024 feature points takes less than 50 m sec on a 1K processor MP-1/MP-2.