Scene image recognition with multi level resolution semantic modeling

摘要: In this paper, we propose a multi-level resolution semantic modeling for automatic scene recognition. The basic idea of the is to classify local image regions into concept classes such as water, sunset, or sky, and use occurrence frequency region's concepts global representation [1]. However, how decide size trial problem. optimized region would be dynamically changing different types. Therefore, paper proposed dynamical (Multi-level resolution) model, fusion probabilities types several resolutions final recognition image. Experimental results show that rate using our algorithm much better than conventional method