Showing papers by "Antonio Torralba published in 1999"

Global semantic classification of scenes using power spectrum templates

Abstract: Scene recognition and content-based procedures are of great interest for image indexing applications processing very large databases. Knowing the context of a scene, a retrieval system may compute its semantic category in advance and filter out scenes belonging to irrelevant classes. In this paper, we introduce a computational approach which classifies and organises real-world scenes along broad semantic axes. Fundamental to our approach is the computation of global spectral templates providing a continuous organisation of scenes between two categories. These templates encode the structure which is discriminant between two categories. We propose a hierarchical procedure of two stages, that organises images along three semantic axis. Firstly, all the scenes are classified according to an Artificial to Natural axis. Then, natural scenes are organised along the Open to Closed axis whereas artificial environments are classified according to the Expanded to Enclosed scenes axis.

Semantic organization of scenes using discriminant structural templates

Abstract: In this paper, we present a procedure for organizing real world scenes along semantic axes. The approach is based on the output energies of linear discriminant filters that take into account, or not, spatial information. We introduce three semantic axes along which pictures are ordered. The main semantic axis computes the degree of naturalness of a scene. Then, urban pictures are evaluated according to their degree of verticalness and natural scenes, according to their degree of openness. We observe the emergence of typical scene categories such as beach, mountain, skyscrapers, city center etc., along the axes.

An efficient neuromorphic analog network for motion estimation

Abstract: Optical flow estimation is a critical mechanism for autonomous mobile robots as it provides a range of useful information. As real-time processing is mandatory in this case, an efficient solution is the use of specific very large scale integration (VLSI) analog circuits. This paper presents a simple and regular architecture based on analog circuits, which implements the entire processing line from photoreceptor to accurate and reliable optical flow estimation. The algorithm we propose, is an energy-based method using a novel wideband velocity-tuned filter which proves to be an efficient alternative to the well-known Gabor filters. Our approach shows that a high level of accuracy can be obtained from a small number of loosely tuned filters. It exhibits similar or improved performance to that of other existing algorithms, but with a much lower complexity.