In the advent of video data explosion, to understand the concept of the video, knowledge of representative data selection and summarization has become essential. In this regard, application of video key frame detectio...
详细信息
ISBN:
(纸本)9789813290884;9789813290877
In the advent of video data explosion, to understand the concept of the video, knowledge of representative data selection and summarization has become essential. In this regard, application of video key frame detection is becoming increasingly critical. Key frame selection of videos is the process of selecting one or more informative frames that depict the essence of the video. In state of the art, researchers have experimented with shot importance measure [1], epitome-based methods [2], and sparsecoding techniques [3] to find informative frames of video. We propose block sparse coding formulation, which exploits the temporal correlation of video frames within the sparsecoding framework for key frames selection. We solved the block sparse coding formulation using the Alternating Direction Method of Multipliers (ADMM) optimization. We show the comparison of results obtained with the proposed method, state-of-the-art algorithm [3] and ground truth on TRECVID 2002 [4] dataset. Comparison results show 8x run time and 6% F-score improvement compared to state of the art.
The transmission rate of uplink and downlink has been increasing with the application of Long Term Evolution (LTE). Correspondingly, for LTE wireless distributed base station, the amount of data transferred between Ba...
详细信息
ISBN:
(纸本)9781479921867
The transmission rate of uplink and downlink has been increasing with the application of Long Term Evolution (LTE). Correspondingly, for LTE wireless distributed base station, the amount of data transferred between Base Band Unit and Radio Remote Units has been growing. But nowadays, the used equipment can only support a limited intelface rate. Aiming at this problem, this paper proposes an adaptive frequency domain floating-point coding (AFFC) for the data of optical fiber interface in-phase and quadrature data (IQ data). In our method, IQ data is firstly converted into frequency domain by fast Fourier Transformation, and we can find its blocksparse distribution;secondly, we use the block floating-point coding to compress IQ data, and use Hoffman code to compress block index. Simulation results show the AFFC can compress the IQ data bits effectively and achieve the purpose of decreasing bandwidth of LTE-Ir intelface. Compared with the existing compression algorithms in time domain, the proposed algorithm could achieve higher compression rate, and smaller magnitude of the error vector.
暂无评论