A Vehicle Flow Detection Algorithm Based on Low-Rank Matrix

The traditional detection method of vehicle flow detection have limitations to low accuracy in the complex scene, this paper proposes a new vehicle flow detection algorithm based on low-rank matrix. The algorithm firstly introduce the Ising model and Robust Principal Component Analysis (RPCA) to get the no-convex energy function, and then employ the singular value decomposition (SVD) and iterate step by step to solve the problem that energy function is non-convex, and then optimize the energy function to detect the foreground vehicles. Finally, we count the number of vehicles by using virtual coil. Compared with the frame-difference method and the mixed Gaussian algorithm, the experimental results show that the proposed method can detect vehicle effectively and accurately, even in fog weather.