Artificial Color Contrast for Machine Vision and Its Effects on Feature Detection

Color information is useful in vision-based feature detection, particularly for applications involving natural objects. One of the factors influencing the success rate of color machine vision in detecting a target is its ability to characterize the color. When unrelated features are very close to the target in the color space, which may not pose a significant problem to an experienced operator, they appear as noise and often results in false detection. This paper describes a method for creating artificial color contrast (ACC) between features in color space with objective of highlighting the target while suppressing surrounding noise; the development of this ACC method has been motivated by the ability of the human to perceive fine gradation of a variety of color especially for natural products where, in most cases, humans are still the sensor of choice. The efficiency of this method is demonstrated on representative automation problems.