Eric PAQUET,¹ Pascuala GARCÍA-MARTÍNEZ^2,* and Ignacio MORENO^2,3

¹Visual Information Technology, Institute for Information Technology, National Reseach Council of Canada, Building M-50, Montreal Road, Ottawa (Ontario), K1A 0R6, Canada, ²Departament Interuniversitari d'Òptica, Facultat de Física, Universitat de València, Dr. Moliner, 50, 46100 Burjassot (València) Spain, ³Departament de Física, Grup d'Òptica, Universitat Autû¢¦oma de Barcelona, Edifici Cc, 08193 Bellaterra (Barcelona) Spain

(Received March 11, 1996; Accepted February 14, 1997)

Correlation methods for recognition of polychromatic range images are presented. First, polychromatic range images are described. The correlation can be performed using either a single channel compressing both the range and the color images or a multichannel approach. The advantages of both methods are discussed. Translation invariant recognition along the x,y, and z axis is obtained. In the single channel approach, the range and the hue are combined using phase and amplitude coding. Then the double-channel approach is considered using sine-coding for the input image and phase-coding for the reference image. Finally, both channels are combined in order to improve the discrimination capability of the system. Optical implementation is performed.

Key words : range images, color optical pattern recognition, phase encoding, amplitude encoding