Advanced Imaging


Advanced Imaging Magazine

Updated: January 12th, 2011 10:01 AM CDT

MV Software Becoming Integral Part of Color Imaging

Component Integration
True color format
Stemmer Imaging
By using a Bayer filter, images can be switched into a true color format.
Intuitive interpretation of colors
By separating the achromatic black and white component (Intensity) from the chromatic components (Saturation and Hue) results in a more intuitive interpretation of colors.
Landscape image
MVTec Software
This merging of seven channels breaks down this landscape image.

By Barry Hochfelder

Euresys’ eVision EasyColor is another suite of tools tailored to color applications. It supports several Intensity/Saturation/Hue color systems by separating the achromatic black and white component (Intensity) from the chromatic components (Saturation and Hue). This allows a more intuitive interpretation of colors and helps to segment them while eliminating lighting effects. It’s necessary, when doing color image processing, to convert the RGB images from the camera to another color space, such as LSH, ISH or YSH.

While also offering traditional color image processing functions like Bayer pattern, EasyColor also provides color analysis functions. Its lookup table, which can convert one color system to another, can be used “on the fly” to avoid storing the transformed image. It’s possible, for instance, to alter the U component of a YUV image while the entire image remains stored in RGB format.

Typical applications for color classification must answer the question: Is this object’s color the right one? MVTec’s HALCON offers a couple of methods to answer the question: A histogram-based analysis that requires no training and no region of interest (ROI). In the trained method, however, an ROI is necessary and the color has to be established off-line, which requires a few seconds.

HALCON offers three different classifiers in one software package: MLP (multi-layer perception), SVM (support vector machines) and GMM (Gaussian mixture method). MLP and SVM need the training of a rejection class that describes the color of the image background. GMM doesn’t.

HALCON also processes images that contain multiple channels, allowing processing of normal color images as well as multi-spectral images like those used in satellite-based remote sensing.

Subscribe to our RSS Feeds