An Algorithm of Fire Image Separation Based on Geometric Independent Component Analysis

In order to solve the problem of detection and identification of fire and disturbance in large space image fire detection, an algorithm of fire moving target recognition based on the theory of Geometric Independent Component Analysis is proposed. And then the feature of detected target is extracted, including area, perimeter and circularity. In our experiments, we compared this algorithm with common used algorithm such as frame difference method and background subtraction algorithm. The results indicated that, compared with the two algorithms, the proposed algorithm can better extract targets in fire images completely. Binding the characteristic parameters of the flame, it can enhance the accuracy of fire detection, besides, the computing time satisfies real-time requirements