Browse Prior Art Database

Digital Method to Eliminate Shade Effects in Images

IP.com Disclosure Number: IPCOM000061839D
Original Publication Date: 1986-Sep-01
Included in the Prior Art Database: 2005-Mar-09
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Wahl, F: AUTHOR

Abstract

In machine vision systems using edge information obtained from an image of an object or scene to be interpreted, shadow silhouettes in such an image are disturbing. This article suggests a procedure for shade elimination. A scene is captured twice by a TV camera, using two light sources located at different places. The resulting two images have different shade artifacts. Relevant edges stay stable in both images. Binary gradient images are then generated, and are compared on the signal level. For this comparison, the images are combined pixel by pixel, e.g., by a logical AND operation. As a result, only edge elements which appear in both binary gradient images survive. Edge elements created on shadow borders are fully eliminated.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 69% of the total text.

Page 1 of 1

Digital Method to Eliminate Shade Effects in Images

In machine vision systems using edge information obtained from an image of an object or scene to be interpreted, shadow silhouettes in such an image are disturbing. This article suggests a procedure for shade elimination. A scene is captured twice by a TV camera, using two light sources located at different places. The resulting two images have different shade artifacts. Relevant edges stay stable in both images. Binary gradient images are then generated, and are compared on the signal level. For this comparison, the images are combined pixel by pixel, e.g., by a logical AND operation. As a result, only edge elements which appear in both binary gradient images survive. Edge elements created on shadow borders are fully eliminated. The computational burden for subsequent image interpretation thus can be kept moderate, as dealing with ambiguous edge information is not necessary. It should be noted that the criterion for surviving edge elements can also be based on non-binary gradient values (e.g., as a monotone function of the gradient values and the correlation of both gradient values). Furthermore, it is obvious that the proposed technique can be extended to a similar approach with more than two light sources. The proposed method can be easily built up by standard off-the- shelf image processing boards controlled by any host system (e.g., a personal computer). The first light source is activated, and the correspo...