Recognition of Moving Objects in Videos of Moving Camera with Harris Attributes

Abstract

Moving object detection is one of the most essential problems in image processing. It attracts many attentions recently. In the paper it is also assumed that the camera is moving. Major part of previous moving car detection methods engages radar signals. For online moving object detection, we suggest to employ hierarchical partitioning over the attributes extracted from image. Each moving object corresponds to a partition. Unlike the traditional partitioning algorithms, the threshold distance in the suggested method is not fixed. This threshold value is tuned by a Gaussian distribution. Harris attributes are applied to capture the corner attributes. Experimentations show the suggested method outperforms other competent methods.