Code readers

An industrial image-based code reader captures an image of a code and then tries to locate and decode the symbol. Modern devices can read: 1D barcodes DataMatrix QR PDF417 DPM codes among others.

A 1D barcode encodes information mainly through the width and arrangement of bars and spaces. Examples: Code 39 Code 128 GS1-128 EAN UPC

A 2D code encodes data in two spatial directions. Examples: DataMatrix QR DotCode This often allows considerably more information to be stored in a small area than in a linear barcode.

Code 128 is a high-density linear barcode that can represent the full ASCII character set through different character sets. It is widely used in industrial and logistics applications.

GS1-128 uses the Code 128 symbology but adds the GS1 data structure with application identifiers. This allows batch, date or product information to be transmitted in a standardized way.

Code 39 is a linear barcode that can represent digits, upper-case letters and selected special characters. It has traditionally been widespread in many industrial applications.

EAN and UPC symbologies are used above all in retail for product identification. They typically carry GTIN-based product identifiers.

PDF417 is a stacked two-dimensional barcode symbology. Several linear code rows are arranged above one another and combined with error correction.

DataMatrix ECC200 is a very widespread 2D matrix symbology. The symbol has characteristic finder and timing structures and uses error correction. Square ECC200 symbols range from 10 × 10 to 144 × 144 modules depending on the variant.

QR is also a two-dimensional matrix code. The striking position detection patterns in three corners are typical. A standard QR code starts at 21 × 21 modules and grows with the version.

Both are matrix codes but use different symbol structures. DataMatrix is used very frequently for industrial marking and small components. QR is widespread in the consumer, document and logistics fields, among others. Which symbology is more suitable depends on the applicable standard, the space, the data content and the process.

DMRE stands for Data Matrix Rectangular Extension. It extends DataMatrix with additional rectangular symbol sizes and therefore suits narrow marking areas, for example. Whether a reader supports DMRE has to be checked explicitly in the datasheet.

A module is the smallest basic unit of a code. With DataMatrix or QR that is typically a single light or dark cell. With 1D barcodes the basic module typically corresponds to the width of the narrowest bar or space element. Zebra accordingly describes the module as the smallest divisible unit of a barcode.

PPM means: pixels per module The figure describes how many camera pixels the smallest code module is rendered across in the camera image.

First: pixels/mm = camera pixels / field of view [mm] Then: PPM = pixels/mm × module size [mm] Or directly: PPM = camera pixels × module size / field of view Example: 1,920 pixels 120 mm field of view 0.25 mm module size 1,920 / 120 = 16 pixels/mm 16 × 0.25 = 4 PPM The code therefore has roughly 4 pixels per module.

The X dimension is the nominal width of the smallest module of a barcode. With a DataMatrix symbol this corresponds to the physical size of one cell.

For the active symbol area: symbol size = number of modules × X dimension Example: 32 × 32 modules X dimension = 0.25 mm 32 × 0.25 = 8 mm The active code area is therefore roughly 8 × 8 mm. For the real printed or marked area, the quiet zone required has to be taken into account as well.

Rearranged from the PPM formula: camera pixels = desired PPM × field of view / module size Example: desired resolution = 4 PPM FOV = 100 mm module = 0.2 mm 4 × 100 / 0.2 = 2,000 pixels At least roughly 2,000 pixels would be required horizontally by calculation.

FOVmax = camera pixels × module size / desired PPM Example: 2,448 pixels 0.2 mm module 4 PPM 2,448 × 0.2 / 4 = 122.4 mm

There is no universal figure. The minimum depends on: reader decoder symbology code quality perspective focus movement lighting One specific Cognex IDMax algorithm, for example, names 4 pixels per element for DataMatrix and QR. Zebra, by contrast, requires at least 10 pixels per module for its ISO-based barcode quality metrics setup. Those are different tasks: reading is not the same as verifying to a standard.

That can be possible with certain readers and very good codes, but it must not be promised across the board. What matters is the minimum resolution of the specific reader or decoder. For a sound design, the manufacturer's datasheet or a test should therefore be used.

A reader primarily has to decode the content correctly. A verifier additionally has to measure quality parameters of the individual modules or bars reliably. That requires a much more precisely defined optical image. Zebra names at least 10 pixels per module for its ISO-compliant quality metrics setup, for example.

DPM means direct part marking. The code is applied directly to the workpiece instead of being printed on a label.

Examples: laser marking dot peen marking engraving etching The surface structure can differ greatly as a result.

DPM codes often have: little brightness contrast directional reflection irregular module shapes textured surfaces That is why the lighting matters particularly.

Low-angle or dark-field lighting is often very effective. The indentations of the dot peen marking scatter the shallow incident light towards the camera.

That depends heavily on the material and the laser process. Possible solutions: coaxial dome dark field diffuse low-angle lighting A practical lighting trial is strongly recommended with DPM.

Code verification assesses the quality of a code against defined standards. It answers not only: “Can my reader read the code?” but, for example: “How robustly and how compliantly was this code produced?”

A reader is optimized to decode data reliably. A verifier measures and assesses symbol quality under defined conditions. A capable reader can therefore read a code that is rated poorly under standards-compliant verification.

ISO/IEC 15415 is a central quality standard for two-dimensional printed matrix codes. Several quality parameters of the symbol are assessed.

ISO/IEC 15416 covers the quality assessment of linear barcodes.

ISO/IEC 29158 covers the quality assessment of directly marked DPM codes and takes their particular surface and marking properties into account.

During verification, a quality rating is derived from various measured parameters. Depending on the standard and the presentation, this is given numerically or as a letter grade, for example. Important: the rating also depends on the standard used and the specified measurement setup.

Symbol contrast describes the difference in brightness between light and dark areas of a symbol. Too little contrast makes both reading and verification harder.

Put simply, modulation assesses how well the local contrast of individual modules is preserved in relation to the overall contrast. Weak or unclean modules can be rated worse as a result.

This parameter describes geometric deviations of the symbol between the axes. Unwanted stretching or compression can worsen the rating.

Grid non-uniformity assesses deviations of the actual module positions from the ideal module grid.

Fixed pattern damage assesses damage to structurally defined areas of the code, for example finder or timing structures.

This value describes how much of the available error correction is still unused. The more the code already relies on error correction, the smaller the remaining reserve.

Decode checks whether the symbol can in principle be decoded correctly with the standard reference decoding procedure.

The quiet zone is a clear area around the code. No disruptive lines, characters or structures should lie there. A quiet zone that is too small or soiled can impair recognition and quality assessment.

2D codes such as DataMatrix and QR contain error correction data. Damaged or illegible parts can therefore be reconstructed up to a certain extent. That does not mean, however, that arbitrarily badly damaged codes stay reliably readable.

The read rate is the proportion of codes read successfully. read rate = successful reads / read attempts × 100 % Example: 9,980 successful reads 10,000 attempts 9,980 / 10,000 × 100 = 99.8 %

There is no generally valid target figure. Whether 99.8 %, 99.99 % or some other value is sufficient depends on: production volume the cost of a no read traceability requirements possible rework In critical traceability processes, even a few no reads can have considerable consequences.

No read means the reader could not decode a valid code within the time or settings available.

Common causes: module too small wrong focus motion blur overexposure underexposure reflections damaged code quiet zone too small perspective wrong lighting too few PPM trigger fault

The commonest cause is motion blur. Other possibilities: wrong trigger point code leaves the depth of field too little light for a short exposure time vibration change in perspective The motion blur should be checked by calculation.

Possible causes: focus range module too small in the image at greater distance too little illuminance limited optical resolution reader outside its specified reading range Manufacturers usually state reading distances for their readers in combination with specific module sizes. KEYENCE, for example, specifies reading ranges depending on the DataMatrix module size.

Many current image-based readers support multi-code applications. How many codes can be read reliably per capture depends on: reader resolution number code size decoding time field of view

There is no general formula. The rate depends on: frame rate exposure time decoder code type number of codes image resolution search area communication among other things. For sound figures, the specific model has to be checked and tested under real conditions.

OCR means optical character recognition. A vision system recognizes characters and converts the image into a text string, for example. Example: image contains: LOT 12584 OCR returns: LOT 12584

OCV means optical character verification. The focus here is not primarily on freely recognizing unknown text, but on checking whether the expected characters are present and printed correctly.

OCR: “What does it say?” OCV: “Does it say what it should say?”

OCR reads visible written characters. Code reading decodes a defined barcode or matrix code symbology. Codes are optimized for machine recognition and therefore normally have considerably more structural redundancy than ordinary text.

There is no universal minimum. The resolution required depends on: font character height contrast print quality OCR algorithm lighting movement The minimum character height of the specific vision system or OCR tool should therefore be used.

Deep learning OCR uses trained neural networks for character recognition. That can offer advantages particularly with variable fonts, difficult backgrounds or fluctuating print quality.

PPM = number of pixels × module size / FOV

pixels = target PPM × FOV / module size

FOVmax = pixels × module size / target PPM

active symbol size = number of modules × X dimension Take the quiet zone into account in addition.

Important parameters: code type printed or DPM module size code size field of view working distance speed number of codes surface PLC/network interface It can then be checked whether the minimum module size and field of view of the specific reader fit.

A DPM reader should not only have a capable decoder but also match the type of marking and surface. Particularly important are: lighting options autofocus or suitable optics module resolution DPM algorithms mechanical design A real sample test is highly recommended for demanding DPM applications.

What matters is: the shortest exposure time possible enough light suitable sensor resolution trigger performance decoding time frame rate Do not look at the nominal “reads per second” figure alone.

The reader has to provide enough: FOV pixel resolution decoding performance process time at the same time. Each individual code still needs sufficient pixels per module within the shared field of view.

The reader or decoder and the application determine the value required.

A readable code does not automatically have a good ISO quality rating.

A stronger decoder cannot fully compensate for a fundamentally unsuitable lighting geometry.