Approximating Content-Based Object-Level Image Retrieval

Object-level image retrieval is an active area of research. Given an image, a human observerdoesnot see randomdots of colors. Rather,he she observesfamiliarobjectsin the image. Therefore, to make image retrieval more user-friendlyand more e ectiveand e cient, object-level image retrieval technique is necessary. Unfortunately, images today are mostly represented as 2D arrays of pixels values. The object-levelsemanticsof the images are not captured. Researcherstry to overcome this problem by attempting to deduce the object-level semantics through additional information such as the motion vectors in the case of video clips. Some success stories have been reported. However, deducing object-level semantics from still images is still a di cult problem. In this paper, we propose a "color-spatial" approach to approximate object-level image retrieval. The color and spatial information of the principle components of an object are estimated. The technique involves three steps: the selec...

• Type: Journal
• Year: 2000
• Where: MTA
• Authors: Wynne Hsu, Tat-Seng Chua, Hung Keng Pung
• Keywords: Images, MTA 2000, Object-level Image Retrieval, Object-level Semantics