摘要

Based on the features of agricultural disease-control video, we determined a correlation between audio and video to segment agricultural disease-control videos. In addition, we constructed a scene semantic model of agricultural disease control and designed a scene-detection method for agricultural disease-control videos. The experimental results showed that scenes of agricultural disease-control videos segmented by existing scene-detection algorithm on semantic unit were incomplete and unavailable. The method employed in this paper overcame the aforementioned defects, thus improving the accuracy of agricultural disease-control video retrieval while simultaneously providing solution to the conflict between video qualities and the costs of the flow of agriculture information service in China's mobile network.

全文