1D (one-dimensional) Barcode: Also known as a linear barcode, a machine-readable code representing data in the widths and spacings of parallel lines. Examples include: Codabar, Code 128, Code 39, EAN, Interleafed 2 of 5, Pharmacode, POSTNET, and UPC.
2D (two-dimensional) barcode: A machine-readable code that stores data both horizontally and vertically. Examples include: Aztec, Data Matrix, MaxiCode, and QR code.
3D vision systems: An optical device used to capture and process three-dimensional images of objects or scenes. By mapping scenes in 3D, these cameras can handle environment changes and variations in analyzed objects. Ideal for inspection, process control, measurement, and robot guidance applications.
30-degree lighting: Illuminates the target at a 30-degree angle. Depending on the variation of the lighting (30Q, 30T, 30S) illumination comes from all four sides, two sides, or a single side. This lighting technique is ideal for verifying direct part mark (DPM) codes on curved, flat, and semi-shiny parts. It's also good for dot peen codes marked on a textured background.
45-degree lighting: Illuminates the target from a 45-degree angle. Incoming light is reflected away and not up towards the camera. This lighting type is used for verifying label-based barcodes and direct part mark (DPM) codes on anodized, plastic, or matte surfaces.
90-degree lighting: Shines straight down and is reflected back up into the camera. This is the only lighting option for verifying direct part mark (DPM) codes on shiny, reflective surfaces.
Algorithm: A set of instructions and calculations that help a computer achieve a goal. In factory automation, algorithms can sort through massive data sets from sensors, scanners, and digital cameras to reveal inefficiencies, boost quality, and recommend improvements. A “learning” algorithm uses trial-and-error and learn-by-example methodologies to optimize production processes without human intervention.
Aperture: For 1D linear codes, aperture refers to the sample size taken across the code's length to read bars and spaces using a scan line, specified in mils. For 2D codes, aperture refers to the circular sample captured at grid intersections. ISO standards dictate aperture sizes for accurate grading.
Application standard: Rules set by an organization that set the guidelines for barcode marking and grading, such as what type of symbology is acceptable, the standard to grade against, the minimum acceptable grade, aperture, x-dimension range, and lighting angles required, and how the data within the barcode must be formatted.
Artificial Intelligence (AI): Computing techniques that mimic human decision-making, using automation to perform tasks that are difficult for humans. AI algorithms are central to factory automation because they can use technologies such as image recognition and natural language processing to reduce human error and anticipate production challenges.
AI hallucination: Instances where an AI system generates incorrect or misleading outputs or responses that are not grounded in reality or factual data – the AI essentially "imagines" data that doesn't exist. This typically occurs in generative models, like language or image generators, which may fabricate information or make false connections, and is rarely an issue in industrial AI systems.
AI bias: AI systems produce errors or unfair outcomes due to biases introduced during the machine learning process. These biases may stem from skewed training data, flawed algorithms, or a lack of diversity in the data set, resulting in unfair or discriminatory outcomes. In manufacturing, AI bias can compromise machine vision systems by distorting inspection results, potentially overlooking or misidentifying defects, which could hurt product quality and operational efficiency. It can also perpetuate societal inequalities, reinforcing discrimination based on race, gender, or other factors.
Area scan: Type of 2D camera used in machine vision. It captures detailed, still images. Better for inspecting smaller, defined items. These can be simple, single-purpose sensors or more full-featured systems.
Axial non-uniformity: Refers to how “out of square” the shape is or the amount of deviation along the 2D code’s horizontal and vertical axes.
Back propagation: Algorithm used to train neural networks by minimizing the error between predicted and actual outputs. It involves a forward pass where inputs are processed to produce an output, followed by a backward pass where the algorithm adjusts the network’s weights (numerical values that determine the strength of connections between neurons) using gradient descent to reduce the error. Repeating this process over multiple iterations helps the model learn and improve accuracy.
Bar width growth: The amount a bar or element size deviates from the ideal size.
Barcode: Machine-readable pattern of parallel lines and spaces (1D) or rectangles, dots, and other geometric patterns (2D) applied to products, packages, or parts that contain data used for informational, marketing, and tracking purposes.
Barcode reader: Device that decodes the data or information in a barcode. Some barcode readers use lasers to decode the information (laser scanners) while others use lights or cameras to capture the barcode image to then read and decode it (image-based barcode readers).
Barcode verification: Process of grading the print quality and readability of 1D, 2D, and DPM codes to specified standards. Barcode verifier Inline or offline device used to grade the quality of a barcode.
Bias mitigation: Techniques and strategies used to reduce or remove bias in AI and machine learning models. This involves finding and addressing biases in training data, algorithms, and decision-making processes to ensure fair outcomes. Common methods include data balancing, algorithmic adjustments, adversarial debiasing, and post-processing corrections.
Bias variance tradeoff: The balance between a model's ability to minimize bias and variance to achieve the best prediction accuracy. High bias can lead to underfitting, where the model is too simple and doesn’t capture the complexity of the data. Conversely, high variance can result in overfitting, where the model is too complex and performs well on training data but poorly on unseen data. The tradeoff involves finding the right level of complexity to minimize total error
Big data: Technologies that pull together enormous data sets from multiple sources, using analytics tools to generate unique insights. Big data lays the foundation of AI because learning algorithms require mountains of information to emulate human decision-making and produce accurate forecasts. An automated factory uses big data analytics to inform its AI initiatives. Technologies that pull together enormous data sets from multiple sources, using analytics tools to generate unique insights. Big data lays the foundation of AI because learning algorithms require mountains of information to emulate human decision-making and produce accurate forecasts. An automated factory uses big data analytics to inform its AI initiatives.
Calibration process: Mapping a camera’s measurement to actual reflectance levels, setting a baseline for verification software reflectance levels. The calibration process adjusts reflectance levels and checks these levels to ensure the most accurate verification results. To be ISO compliant, a verifier must have a calibration process.
Calibration card: Card with master symbols that are graded against a NIST traceable judge card. These symbols can be used to calibrate and check your calibration values.
Cell contrast: ISO/IEC TR 29158 graded parameter that is the DPM version of the ISO/ IEC 15415 parameter called "symbol contrast" that calculates the relative contrast value between bars and spaces.
Cell modulation: ISO/IEC TR 29158 graded parameter that is the DPM version of the modulation that is calculated by grading scale range set to mean of distribution rather than the mean of the maximum and minimum reflectance used in ISO/IEC 15415.
Chemical etching: A form of direct part marking; the subtractive manufacturing process of using baths of temperature-regulated etching chemicals to remove material to create an object with the desired shape.
Classification: Separating data into distinct categories that improve AI effectiveness. In factory automation, data-gathering devices scan the production environment and recognize specific objects like machine parts and electronic components. AI algorithms help classify these objects and route them to their proper destinations in the production line.
Clocking pattern: Provides a count of the number of rows and columns in a 2D code.
Clustering: Segregating a population of data points into clusters to figure out which ones are similar and which ones differ. Learning algorithms scan clusters to interpret their meaning and predict a proper course of action. For instance, clustering algorithms process data from optical scanning devices to decide what to do when a scan detects a product flaw.
Computer vision: Enables machines to interpret and understand visual information from the world. It involves teaching computers to analyze and make decisions based on visual data, such as images and videos, often focusing on finding and classifying objects. Beyond industrial applications, computer vision is used in diverse fields like facial recognition, autonomous vehicles, and medical imaging.
Conformance calibration card: A card with test symbols with intentional imperfections that check the reporting capabilities of the verifier and document conformance to industry standards such as ISO 15415, ISO 15416, and GS1 specifications.
Data cleaning/scrubbing: Removing flawed data to sharpen the effectiveness of a learning algorithm. AI depends on sophisticated data models that must have accurate, reliable, and consistent data sources. Unfortunately, data gathering often yields redundant, inaccurate, out-of-date, or otherwise useless information. Cleaning (or scrubbing) data reduces risks and enhances AI accuracy.
Deep learning: An AI technology designed to automate complex and highly customized applications. Processing takes place via a graphics processing unit (GPU), which enables users to build sophisticated neural networks from large, detailed image sets (hundreds to thousands of images). Leveraging these neural networks, deep learning quickly and efficiently analyzes vast image sets to detect subtle, variable defects and differentiate between acceptable and unacceptable anomalies.
Data Matrix: A 2D code that can encode large amounts of data (up to 2,335 alphanumeric or 3,116 numerical characters) and use an error-correction system to read codes that are as much as 40% damaged. It is made up of black and white cells in a square or rectangular pattern and includes a finder and clocking pattern.
Decodability: Grades the accuracy of bar and space widths compared to their ideal size. A barcode with bar width growth or distortion will score a low decodability grade.
Decode: Checks to see whether the code can be read using the standard reference decode algorithm.
Depth of field: Refers to the distance between the nearest and farthest objects that appear sharp in an image, or the distance from which a barcode reader can successfully decode a barcode.
Direct part marking (DPM): A process to permanently mark parts with product information including serial numbers, part numbers, date codes, and barcodes. This is done to allow the tracking of parts through their full life cycle. Examples include chemical etching, dot peen, and laser marking.
Discriminative AI: AI models that predict outcomes based on given data. These outcomes can be classifications, like whether an image is "good" or "bad" for inspection tasks, or identifying a text paragraph as positive or negative. The predictions are not open-ended; the AI models know what to predict, making this approach typical for machine vision applications.
Dome lighting: Provides uniform light from various angles which results in no glare, even on mirrored objects. It is used most often to inspect shiny, curved, or bumpy surfaces. It can also be used to verify DPM codes printed on curved or textured parts. To be effective, dome lights require proximity to the target.
Dot peen: A direct part marking method composed of a carbide or diamond stylus that rapidly actuates and makes a series of small dots on a material’s surface. The stylus moves along the surface and forms marks including alphanumeric characters and machine-readable codes and logos.
Edge contrast: Measures the difference between adjacent bars and spaces.
Edge count: Pass or fail parameter that counts the number of edges in a 1D code.
Edge learning: AI technology designed for ease of use, processing data on-device, or “at the edge,” with pre-trained algorithms. It’s easy to set up, requiring smaller image sets (as few as 5 to 10 images) and shorter training times than traditional deep learning solutions. No domain expertise is needed – non-vision experts can easily train edge learning tools and generate inspection results in just minutes.
Element: An individual bar or space within a linear barcode.
Ethical AI: The development of AI systems that prioritize fairness, transparency, accountability, and respect for human rights. It ensures AI technologies are deployed responsibly, without causing harm or discrimination, and promotes decision-making that benefits society while safeguarding privacy and diversity. In manufacturing, Ethical AI guarantees that automation and AI-driven processes foster safer and more efficient production while minimizing bias and promoting inclusivity.
Explainable AI (XAI): AI systems designed to clarify their decision-making process, so operators can trust and understand how automation technologies are influencing production. By providing transparency into how AI models arrive at their conclusions, XAI helps manufacturers find potential issues, improve efficiency, and maintain accountability in critical processes. This promotes confidence in automated systems, ensuring more reliable and informed analysis.
Federated learning: A training technique for machine learning models on decentralized data, where the data is distributed across multiple devices or nodes (like smartphones, IoT devices, edge devices, etc.). Instead of centralizing the data and training the model in a single location, the model is trained locally on each device, and the updates are aggregated and shared with a central server.
Field of view: The overall image size or coverage area a camera supports.
Finder pattern: Also called a “locator” or “L” pattern, it connects the left and bottom lines of a 2D code and helps the reader to locate and determine the orientation of a machine-readable code.
Fixed pattern damage: Errors with the finder pattern, clock pattern, or quiet zone of the code.
Generative AI: A subset of AI focused on creating new content, such as text, images, music, audio, and video, by learning patterns from existing data. These models use techniques like deep learning and neural networks to generate original outputs based on the data they were trained on. Unlike industrial AI, which is often used for classification or prediction, generative AI focuses on producing novel, creative content that can mimic human-created content.
Gradient descent: A fundamental optimization algorithm used in machine learning to minimize the cost function by adjusting the model's parameters. It works by calculating the gradient, or slope, of the cost function and moving in the direction that reduces the error, eventually converging to the best solution.
Grid non-uniformity: Measures the largest deviation from the code grid or how much the x and y axes are not at a 90-degree angle.
Global Standards One (GS1): Develops the global standards for business communication, most notably the barcode.
Global threshold: The defining point on a scale from dark to light that determines if a cell is closer to light or dark.
Image recognition: Using AI to analyze the content of a digital image. Image recognition algorithms scan millions of pixels and apply edge learning or deep learning to distinguish between people, places, and things. These technologies enable bar-code scanning and automated inspections that can dramatically improve the productivity of an automated factory. See also: machine vision.
Inline barcode verifier: ISO-compliant device installed in a fixed position on a production line used to automatically grade the quality of a barcode.
International Electrotechnical Commission (IEC): Organization that sets the international standards for all fields of electrotechnology and works in partnership with the International Organization of Standardization (ISO) on some quality standards.
International Organization for Standardization (ISO): Standard-setting body composed of representatives from various national standards organizations. There are hundreds of ISO standards, some of which pertain to barcodes—how the data is put in a barcode, how to decode it, how to measure the size of the barcode, and more. There are three ISO standards that pertain to quality: ISO/IEC 15416 for 1D barcodes, ISO/IEC 15415 for 2D label-based barcodes, and ISO/IEC TR 29158 for 2D DPM codes.
Internet of Things (IoT): A network of sensors connected to the Internet to gather real-time intelligence from machines and human behavior. Devices include smartphones, wearables, video cameras, traffic monitors, and sensors within machinery and production lines. IoT sensors generate the massive quantities of data needed to enable machine learning.
ISO/IEC 15416: Barcode verification standard for 1D barcodes. This standard requires 10 individual scan lines to be taken throughout a code and that a grade is assigned to each line based on the lowest graded parameter for that line. The formal grade is formulated by taking the average of all the individual scan line grades.
ISO/IEC 15415: Barcode verification standard for 2D label-based codes. This standard grades according to eight different image and data-related parameters. The lowest individual grade becomes the overall grade for the code.
ISO/IEC TR 29158: Barcode verification standard for direct part mark (DPM) codes. This standard is an extension of the symbol quality grading methodology defined in ISO/IEC 15415. Additionally, it defines alternative illumination conditions, a few new parameters, modifications to the measurement, grading of certain parameters, and the reporting of the grading results.
Large language models (LLMs): Advanced machine learning models trained on vast amounts of text data to understand, generate, and manipulate human language with high proficiency. Using deep learning and neural networks, they predict and produce text that mimics natural language patterns, enabling automation of tasks like translation, summarization, and content generation. LLMs, such as OpenAI’s GPT, enhance natural language processing applications, driving efficiency and innovation in manufacturing and beyond.
Laser marking: A direct part marking method that labels materials with a laser beam by engraving, removing, staining, annealing, and foaming an object’s surface.
Line scan: Used in continuous inspection applications such as web manufacturing. They take a wide but very thin image as material moves past the scan area. Use software to reconstruct the image line by line. Faster in these applications than standard 2D cameras.
Machine learning: Computing processes that can improve outcomes without human programming. Machine learning algorithms train a computer to seek success and avoid failure millions of times to generate learning outcomes. Image recognition, for example, analyzes millions of images to discern between different objects in a digital picture. See also: training data.
Machine vision: Application of computer vision in industrial settings, using cameras, sensors, and software to inspect, measure, and make decisions about products on manufacturing lines. AI enhances traditional rule-based machine vision by enabling image-based analysis, where systems compare images against reference databases to identify defects or anomalies. This process automates complex visual inspections, improving product quality, reducing errors, and lowering costs. Two leading AI technologies – edge learning and deep learning – help to further simplify automation of highly variable tasks that are too complicated and time-consuming to program with rule-based algorithms.
Machine vision systems: A solution that enables industrial machines to "see". Using cameras and image processing software, these systems capture and analyze two-dimensional images to perform tasks in production, such as inspection, measurement, and identification.
Mil: One thousandth of an inch (.001”).
Minimum reflectance: For a 1D barcode, this parameter checks whether the bars of the code are dark enough to meet the required ratio of light that is reflected off the spaces. For DPM codes, it measures the brightness of the original image. The ISO/IEC TR 29158 standard calls for verification software to automatically brighten an image before conducting the verification process. If the symbol has lower than 5% brightness before the adjustment, it will fail.
Minimum x-dimension: Refers to the limitation of a barcode verifier’s camera resolution. When selecting a verifier, the minimum x-dimension must be smaller than the size of the smallest bar or module in the smallest code printed or marked.
Modulation: Grade based on the amount of variability in reflectivity of the modules. A multi-step process is used to determine the grade.
Modular Vision Tunnels: Flexible scan tunnels designed to boost the speed and efficiency in logistics operations – like sortation, inbound receiving, and outbound logistics. They offer 1- to 6-sided barcode reading, are scalable for many speeds and package gaps, and feature DataMan image-based barcode readers with reliable read rates and mixed symbology decoding.
Multimodal AI: AI systems that process and integrate multiple data types, such as text, images, audio, and video simultaneously, for more accurate and context-aware insights. By combining these inputs, it enhances applications like automated inspections in manufacturing, where both visual and textual data improve defect detection. This capability allows AI to interpret complex environments more effectively, boosting efficiency, precision and greater adaptability across industrial and automation tasks.
Natural language processing (NLP): Branch of AI that enables computers to understand, interpret, generate, and respond to human language. By combining linguistics and machine learning, NLP powers applications like sentiment analysis, language translation, and automated text processing. In, it enhances efficiency through voice-activated control systems, allowing operators to interact with machinery via spoken commands. It also automates the analysis of maintenance logs and reports, finding patterns for predictive maintenance and reducing downtime. NLP can optimize supply chain management by processing procurement orders and shipment data, leading to smarter decision-making and improved operational efficiency.
Neural networks: Algorithms and data nodes inspired by the neural pathways of the human nervous system. A neural network emulates how nerve cells receive information, process it, and send it to other nerves. Thousands of processing nodes in a neural network find patterns in data such as images, text, and spoken words. These patterns help the algorithm decide what to do next and teach itself to perform better every time it succeeds at a task.
Offline barcode verifier: ISO-compliant handheld or desktop device used to manually grade the quality of a barcode, often with batch sample testing.
Optical character recognition (OCR): Translating printed numbers, letters, and characters into meaningful digital data. AI technologies provide a quantum leap in OCR accuracy, enabling scanning of labels that are bent, torn, damaged, and obscured by plastic wrappers.
Overfitting: When a machine learning model is excessively complex and learns not only the underlying patterns in the data but also the noise or random fluctuations. As a result, the model performs very well on the training data but poorly on new, unseen data because it has essentially memorized the training set rather than generalizing from it. This diminishes the model's ability to make accurate predictions on real-world data.
Presence/absence: The process of determining whether a specific part, object, or feature is present or absent in an image. Various software tools, such as Pixel Count, Pattern, and Edge Presence/Absence tools, are used to find parts based on pixel count, pattern recognition, or linear edge features.
Quiet zones: A blank space around a barcode that separates the barcode from neighboring graphics, shapes, and textures. Individual barcode specifications define the amount of quiet zone.
Reflectance margin: Modulation and reflectance margin are often the same, differing only when some modules are determined to be the wrong color and error correction is used.
Regularization: A machine learning technique that helps improve model accuracy by preventing overfitting. It does this by adding a penalty for complexity, which encourages simpler models that perform better on new data. Common types of regularization include Lasso, Ridge, and Elastic Net, each helping to build more reliable models
Reinforcement learning: Type of machine learning where an agent learns to make decisions by interacting with an environment to maximize cumulative rewards. The agent, which could be a software program or a physical robot, receives feedback through rewards or penalties based on its actions. Through trial and error, it refines its strategy, improving performance over time. This approach is especially effective in sequential decision-making scenarios, such as robotics, gaming, or autonomous vehicles.
Robotic process automation (RPA): Devices, software, and processes that automate production, improving productivity and freeing people from tedious, repetitive tasks. RPA devices include robot arms that do tasks like welding automotive parts or assembling microprocessors. RPA software can include algorithms and applications that automate processes that previously required human input.
Scan reflectance profile (SRP): Represents the changing reflectance values of a linear barcode across its width. This forms a basis for the measuring and grading of the quality parameters used.
Sentiment analysis: Process of computationally determining whether a piece of writing is positive, negative, or neutral. It classifies text based on the mood or mentality expressed, using various Natural Language Processing algorithms such as rule-based, automatic, and hybrid approaches.
Symbol contrast: Graded value that measures the difference between the maximum reflectivity in the lightest space and the minimum reflectivity in the darkest bar of a code.
Structured Data: One framework that helps machines understand the meaning of data and process it accurately. Structured data is stored in consistent formats, much like the rows, columns, and cells of a spreadsheet.
Unstructured Data: Another framework that helps machines understand what data means and how it should be processed. Unstructured data is like the text in a word processing file or the pixels in a video. AI algorithms allow factory automation systems to extract knowledge from both kinds of data.
Supervised learning: A type of machine learning where the model is trained on labeled data, with each input paired with the correct output. The model learns to map inputs to the correct outputs by finding patterns in the data during training. Once trained, the model can make predictions on new, unseen data based on these learned patterns.
Swarm intelligence: A collection of algorithms working together to produce intelligence that exceeds the ability of individuals within the swarm. Modeled on the behavior of bees, fish, birds, and other social organisms, a swarm intelligence system assigns simple, specific tasks to individual machine-learning algorithms. Allowing these algorithms to work together can produce learning outcomes far more sophisticated than a single algorithm could generate. See also: neural networks.
Tokenization: Fundamental process in Natural Language Processing (NLP) that involves breaking down a stream of text into smaller units called tokens. These tokens can range from individual characters to full words or phrases, making it easier for machines to analyze and understand human language.
Traceability: Ability to track and trace the history, application, or location of an item through the supply chain. It gathers information about the manufacturing process, like where a product originated, what materials it contains, and how it was manufactured.
Training Data: Digital information that fuels learning in an AI algorithm. AI and neural network algorithms must know the difference between good and bad data. Training data lets them figure it out. In an automated factory, for instance, digital images provide training data that help algorithms learn to find flawed machine parts and keep them out of finished products.
Transfer learning: A machine learning technique where a model trained on one task is repurposed as the foundation for a second task. This approach works best when the second task is related to the first task, or when data for the second task is limited. By using learned features from the first task, the model can adapt more efficiently to a new task.
Underfitting: When a machine learning model is too simple to capture the underlying patterns in the data. This typically happens when the model doesn’t have enough complexity, or the features it uses are too limited. This results in poor performance on both training and unseen data, hindering the model's ability to generalize and make accurate predictions. It’s one of the issues to avoid when building AI models, as it limits the model’s ability to generalize and make accurate predictions.
Unused error correction: Percentage of error correction capability that is available for further incorrect modules.
Unsupervised learning: A machine learning approach where the model is trained on data without labeled outcomes. The goal is for the model to identify patterns, relationships, or structures within the data autonomously. It is commonly used for clustering, anomaly detection, or discovering hidden insights from large datasets.
Validation: Confirms that the format of the barcode data is correct in accordance with application requirements so that a computer system that receives the barcode data can correctly interpret the meaning of that data.
X-dimension: Size of the narrowest bar or smallest module in any barcode; normally referenced in mils.