Image data processing extracts results

Based on vision-system application requirements, suitable processing of imaging data usually yields the desired results. Whether hardware, software, or both are implemented, the data can be analyzed, changed, classified, corrected, enhanced, explored, interpreted, and even modified quickly and easily.

Image data processing extracts results

George Kotelly Executive Editor

georgek@pennwell.com

Based on vision-system application requirements, suitable processing of imaging data usually yields the desired results. Whether hardware, software, or both are implemented, the data can be analyzed, changed, classified, corrected, enhanced, explored, interpreted, and even modified quickly and easily.

Even an industry such as wood manufacturing is benefiting. According to contributing editor Shari Worthington, a sawmill has integrated off-the-shelf cameras, parallel processors, and personal computers into a vision system that inspects, measures, and cuts trees into boards using color bit-mapped images. The system scans boards, processes the image data, and controls the rip-saw settings for more cost-efficient wood production (see p. 30).

A Canadian government agency is analyzing and classifying raw digital data obtained from the LANDSAT satellite to derive a geographical imaging data set of the scanned landscape. By using a PC-based image-processing system, reports contributing editor Larry Curran, the agency is attempting to characterize the derived terrestrial features for ecosystem planning and management (see p. 26).

By implementing the fast Fourier transform (FFT) in silicon for higher data-processing speed, digital FFT integrated circuits (ICs) are challenging the use of optical correlators in complex image-recognition or pattern-matching applications. According to editor-at-large Andy Wilson, several companies are manufacturing single-chip block or floating-point ICs with pipelined multiplier-accumulator stages to accelerate both FFT and inverse-FFT data computational tasks (see p. 36).

Frame-grabber boards serve as the main vehicle for capturing image data. Because more than 50 manufacturers offer PCI-based frame-grabber boards, system integrators must thoroughly evaluate these boards for camera interfaces, image-processing capabilities, and display resolutions to establish the proper performance versus price. In addition, says Andy Wilson, they must ensure that the software developed around their selected board can be easily upgraded (see p. 46).

In Part 2 of our coverage of image-enhancement algorithms, Peter Eggleston explains that choosing the correct imaging techniques is aided by a thorough understanding of their effects and limitations. He cautions that engineers must be aware of artifacts that enhancement functions might introduce, as they could markedly affect later image- and data-processing steps (see p. 21).

More in Boards & Software