Search

We propose a method that can detect pedestrians in a single image based on the combination of Adaboost learning with a local histogram features. Besides, instead of using the raw image for further processing, we introduce a layer enhanced by orientation filters which are superimposed to the original image. Experimental results obtained using the INRIA dataset show the superior performance of our method and thus demonstrate its robustness with the novel enhanced layer embedded pedestrian detector. [ABSTRACT FROM AUTHOR]

Due to the tremendous growth in the number of digital videos, the development of video retrieval algorithms that can perform efficient and effective retrieval task is indispensable. In this paper, we propose a high-level motion-pattern descriptor, temporal motion intensity of moving blobs (MIMB) moments, which exploits spatial and temporal features to characterize video sequences in a semantics-based manner. The Discrete Cosine Transform (DCT) is applied to convert the high-level features from the time domain to the frequency domain. The energy concentration property of DCT allows us to use only a few DCT coefficients to precisely capture the variations of moving blobs. Compared to the motion activity descriptors, RLD and SAH in MPEG-7, the proposed descriptor yield 41% and 20% average performance gains over RLD and SAH, respectively. Having the efficient scheme for video representation, one can perform video retrieval in an accurate and efficient way. [ABSTRACT FROM AUTHOR]