Face recognition search engine that uses distributed computing-based ASM

A Distributed Computing-based Active Shape Model (DCASM) is proposed in this paper. This model analysis the design and construction of search engine for face recognition in large data or high throughput environment. The improvement of DCASM focuses on the following three aspects. First, the new model covers facial feature points of merely half a face and segments face images by parallel lines to construct gray-level appearance models. Second, Locality Sensitive Hashing (LSH) is chosen to build indexes for eye and facial vectors which can significantly speed up searching and matching. At last, we develop a Distributed File System (DFS) with three nodes based on Hadoop theory. For performance evaluation, the result of the comparison proves that DCASM can achieve a better performance.