Robust Moving Targets Detection Based on Multiple Features

Moving targets detection is the most basic part of intelligent video analysis, and its detection effect is directly related to the accuracy of subsequent processing. Aim to cope with the challenges of multi-modal background in complex video environment, the paper proposes a new detection method that combines pixel-based feature and region-based feature. It proposes a new region textural statistic by fuzzy clustering to statistical texture and then fuses it with intensity of pixel. Feature vectors which consisted of rich information are used in background model. Optimal threshold segmentation method is used to obtain adaptive threshold for foreground detection. The experiments indicate that the method can achieve expected results and obviously outperform performance in scenes including multi-modal background.