Innovative stereo images automate inspection of complicated surfaces

Many industrial surface-inspection tasks are undertaken manually by staff involved in costly, tedious, and time-consuming operations. This absence of automation can be attributed to the fact that existing inspection methods that are unable to cope with a variety of products and defects. But the continuous reduction in computing costs clearly indicates the cost-effective advantages of automated industrial-inspection systems in reducing labor efforts, eliminating subjective judgments, and generat

May 1st, 1999

Innovative stereo images automate inspection of complicated surfaces

Many industrial surface-inspection tasks are undertaken manually by staff involved in costly, tedious, and time-consuming operations. This absence of automation can be attributed to the fact that existing inspection methods that are unable to cope with a variety of products and defects. But the continuous reduction in computing costs clearly indicates the cost-effective advantages of automated industrial-inspection systems in reducing labor efforts, eliminating subjective judgments, and generating timely statistical product data.

In pursuit of cost-effectiveness, researchers at the Faculty of Engineering of the University of the West of England (Bristol, England) have developed an innovative technique designed to automate surface inspection. According to Melvyn Smith, this research has resulted in the quality inspection of manufactured components containing intricate or complicated surfaces for which conventional machine-vision techniques have proven inadequate. This research covers the detection of surface defects such as scratches, erroneous indentations, and protrusions found on smooth and textured surfaces. Typical applications include the inspection of decorative ceramic wall tiles, china products, stone surfaces, and painted parts.

In operation, the novel machine-vision system captures three separate images of a fixed part under a sequence of controlled illumination conditions. Local intensity data within each image are then used to derive an array of surface-normal vectors across the part. Then, the system removes any coincident surface chromatic patterns to reveal underlying surface flaws. To classify and quantify the isolated three-dimensional (3-D) surface aberration, a domain-mapping technique is performed and the recovered surface is displayed in relief.

This new vision approach also has application for the inspection of complex topological features that might be obscured by a coincident colored pattern. "A particularly difficult surface-inspection problem is that of inspecting a nonplanar surface with a pseudorandom chromatic pattern," says Smith. "Such surfaces incorporate a regular 3-D topographic surface relief pattern associated with a pseudorandom test pattern," he adds.

During inspection, the integrity of the concealed topographic form must be ensured. In such tests, the chromatic pattern obscures the topographic pattern, making inspection using conventional image analysis highly problematic. However, using this new technique, surface coloring can be separated from the obscured surface topography.

More in Boards & Software