New standard improves verification of Data Matrix codes

When creating any standard, certain “rules” apply. A standard must incorporate some features critical to every company that adopts it, but inevitably, every company will identify critical features that the standard lacks. As a result, standards must evolve as relevant information changes and as user experiences reveal their shortcomings. The standards for verifying the Data Matrix codes that are used to identify parts for inspection illustrate these principles perfectly.

Currently, three standards are used to determine the mark quality of Data Matrix codes—ISO 16022, ISO 15415, and AS9132 (Refs. 1–3). ISO 16022 is the 2-D symbology specification. It describes how to construct a Data Matrix code, and it guides developers in generating software algorithms that read the code itself. The initial release also included an appendix that defined mark quality metrics.

Essentially, the standard covers how to encode the information, how to decode it when you read it, and how to verify that it is correct. The standard assumed, however, that codes consisted of black marks applied to white labels (rather than directly to a part), and it included no measurable quality standards.

When a product is destined for military or other high-reliability applications, manufacturers need criteria for ensuring that a Data Matrix mark will be readable throughout the product’s life—heat, friction, and other indignities notwithstanding. ISO 15415 was developed with the intent to provide a verification standard for demanding applications. Like 16022, this standard assumes that all images consist of black marks on white paper labels. It includes only a single lighting configuration, and the calibration routine limits the system to a fixed exposure and fixed gain for image analysis. If you adhere to the requirements, you will find that many codes that an ordinary person can read easily will fail an automatic verification system.

Fig. 1 a) The old verification standard expected black marks on white paper, producing matched peaks near 0 and 255 on the 8-bit gray scale. b) Marking directly on the part produces marks that peak a considerable distance from 0 and 255, making verification more difficult. Courtesy of Cognex.

Yet, as companies increasingly turn to using direct part marking (DPM), the appearance of marks is becoming much less uniform. Data Matrix marks are imprinted using a variety of methods, so their color and shape can deviate considerably, and the variety of surfaces and materials on which they are imprinted rarely provides a uniform white background. Figure 1b shows a more realistic high-contrast hologram, where the mark is less than black, the background is other than white, and the peaks are not equal.

Although many companies follow ISO 15415 for paper-based Data Matrix codes, companies that use dot-peen, laser, or electrochemical-etch marks have to conform to AS9132. Created by the SAE’s International Aerospace Quality Group, this document describes how to make the marks. It was never intended to address camera-based verification of a mark. It outlines no measurement method, no lighting configuration, and no assurance of consistency in the way that companies implement its provisions.

Carl Gerst, director of ID products for Cognex, noted, “When people mark parts by laser-etching, dark elements tend to be larger than light elements. CO2 lasers will 'over burn’ during the marking process. According to AS9132, overmarking dark cells by as little as 5% will produce a verification failure.”

Because of the shortcomings of the early standards, Cognex and other imaging companies, working with the Association for Automatic Identification and Mobility (AIM), have prepared a new specification, which the Department of Defense intends to adopt. It “explains how to both specify and report quality grades in a manner complementary to, yet distinct from, the method in ISO/IEC 15415.” (Ref. 4.)

The new Quality Guideline is more flexible and realistic than ISO 15415, as its specifications are independent of the marking method. Whereas 15415 permitted only one lighting configuration, the new standard specifies four:

• a diffuse perpendicular (on-axis/bright-field) setup, which illuminates the symbol with diffuse light incident at a 90° angle;
• a diffuse off-axis setup, which illuminates the object from below and is effective on curved surfaces;
• a low-angle, two-direction technique that aims light at the part at an angle of 30° ±3° from two sides, illuminating the symbol area with uniform energy; and
• a low-angle, four-direction technique in which the center of the beams from opposing pairs of lights are coplanar, and the planes remain at right angles to each other.

a)		b)
Fig. 2 a) Marks that would fail ISO 15415 verification will produce much better results using b) the techniques defined in the new AIM standard. Courtesy of Cognex.

Using DPM to identify products undergoing inspection represents a considerable improvement over paper labels that can become defaced or can fall off. The implementation of the new AIM DPM Quality Guideline should help manufacturers improve their ability to apply Data Matrix marks in a consistent method and make verification of the marks both easier and more reliable.