Pictorial Recognition of Objects

This paper describes an eecient approach to pose invariant pictorial object recognition employing spectral signatures of image patches that correspond to object surfaces which are roughly planar. Based on Singular Value Decomposition (SVD), the aane transform is decomposed into slant, tilt, swing, scale and 2D translation. Unlike previous log-polar representations which were not invariant to slant (i.e. foreshortening only in one direction), our log-log sampling connguration in the frequency domain yields complete aane invariance. The images are preprocessed by a novel model based segmentation scheme that detects and segments objects that are aane-similar to members of a model set of basic geometric shapes. The segmented objects are then recognized by their signatures using multi-dimensional indexing in a pictorial dataset represented in the frequency domain. Experimental results with a dataset of 26 models show 100% recognition rates in a wide range of 3D pose parameters and imaging degradations: 0 ?360 swing and tilt, 0 ?82 of slant (more than 1:7 foreshort-ening), more than 3 octaves in scale change, window-limited translation, high noise levels (0 dB) and signiicantly reduced resolution (1:5).