A contemporary approach for object recognition based on spatial layout and low level features’ integration

Abstract

The insightful effort of this paper to make an impulsive and unhampered an interface between the physical and virtual world. In general, objects tracked and recognized while objects placed on surface display. Spatial relation or position of object help to detect the target objects in the complex scene. In this paper, our approach to getting the valuable solution of spatial moments with respect to features like texture, shape and color for object recognition. However, a spatial position with other low- level features’ layout and the chance of probability can be defined according to situation the environment. The contemporary complex databases included occluded and unstructured random natural scenes. According to this paper proposed a new method to integrate the spatial layout with primitive features for object recognition and measure accuracy rate. Our method achieved 78% mean average precision image matching algorithms on Flickr dataset. The present experimental analysis gives low computing cost, medium descriptive power and high image matching features as compared to other existing objects recognition approaches.