The Application Of Video Abstraction Technology In Web Of Things

In recent years, Web of Things (WoT) developed very rapidly, and received extensive attention of the society and academic circles. WoT brings huge market prospects and intelligent, networked thinking to open video technology. However, over time, the monitoring image and video goes through a linear growth and brings huge human cost to visitors. Based on the above problems, we introduced video summarization technology. Through the analysis of the structure and content of video, extract meaningful information from the original video, and recompose the meaningful information, condensed into video semantic content of the full performance.In the video summarization and retrieval system, key frame extraction is an important process. There are existing key frame extraction methods-k-means clustering algorithm, the AP clustering algorithm, etc. However, the most widely applied k-means algorithm is not perfect because it has the need to think about "k’values, which is not satisfactory in adaptability and extract effect of key frames of video.Under the above background, in this paper, the key frame extraction technology of the video summary was studied. The research work and contributions include the following aspects:1. In this paper, a similarity measure algorithm based on nonlinear fusion algorithm of color and texture feature is proposed. And take the video frame fusion measure as AP clustering similarity matrix, which is applied to the video in algorithm.2. In this paper, aiming at the shortcomings of the traditional AP clustering algorithm, AP clustering algorithm based on dynamic damping factor is put forward, and applied to the static video abstract algorithm. Simulation results are consistent with expectations, which improve the quality of the extracted key frames.3. The static video evaluation algorithm CUS[7] is improved for the first time. Through introducing the key frames for clustering and user parameter redundancy rate, make it more scientifically reflect the effect of static video abstract algorithm. Algorithm results described in this article and VSUMM[7], STIMO[18] algorithm are analyzed and compared to verify the accuracy and efficiency of the algorithm described in this article.4. In this paper, the design and realization, of the intelligent video open system is finished. And the algorithm of this paper is applied in it to realize the intelligence of the video open platform.