Focused color intersection with efficient searching for object detection and image retrieval

Abstract

Similarity of color histograms is an important cue for detecting colored objects in complex scenes. It is employed in several applications such as image retrieval, object detection, tracking, etc. Focused color intersection efficiently matches the model against parts of the scene using histogram intersection. We propose an upper bound pruning technique which increases the efficiency of searching for a match by concentrating on more promising regions in the image.