✕
When The Imaging Source set out to implement a high-precision optical character recognition (OCR) system for industrial imaging, applications engineer Tommie Miner turned to Theia Technologies’ ML610M variable focus lens, The Imaging Source DFK 33UX178 camera, and Zebra’s Aurora Vision Studio software platform in combination with deep learning OCR tools. The objective was to provide fast and accurate recognition of small text at a close working distance of approximately 9 inches (approximately 23 cm), even in demanding industrial environments. Reliable OCR is critical for applications such as product serialization, label verification, electronics inspection, and automated traceability, where even the slightest misreading can disrupt production, compliance, and quality control.
Central to the system’s success was Theia’s ML610M 2/3-inch format varifocal lens, designed with a 1.55 μm pixel size and resolving details with up to 300 line pairs of contrast per millimeter. This extremely high resolution ensured that the fine strokes of small, densely printed text remained sharp and distinguishable when projected onto the camera sensor. In the OCR task, maintaining edge contrast at the smallest letter scale was important because the difference between a “1” and a lowercase “l” or a “5” and an “S” determined whether the code was read correctly. The lens’ true 4K, 12-megapixel optical design, combined with aerospace-standard swept vibration immunity from 20 Hz to 200 Hz to 20 Hz at 10 G (30 minutes per axis at 10 G) and IR correction from 440 to 940 nm, ensures clear images even under a variety of lighting and industrial operating conditions (Figure 1).
Combining pixel density and deep learning for industrial OCR
The DFK 33UX178 camera features a 2/3 inch format, 6.3 megapixel sensor with 2.4 µm pixels, strongly matching the ML610M’s Ø11 mm image circle and 300 lp/mm optical performance. The lens itself has a resolution of up to 300 lp/mm, but the effective resolution of the system is limited by the camera to 200 lp/mm. Even at this level, this pairing delivers extremely sharp, edge-to-edge detail across the entire sensor. The 9-inch working distance produces a dense, high-contrast pixel grid that captures even the thinnest strokes with minimal blur.
This rich pixel data directly supported Zebra’s Aurora Vision Studio deep learning OCR tool. This tool relies on high-quality input to identify characters in industrial applications such as label inspection, product serialization, and electronic component validation.
Aurora’s DL_ReadCharacters filter evaluates both localization and classification probabilities, and when fed sharp, high-contrast edges from 200 lp/mm captures, the network maintains a strong confidence score (OcrResult.Score) and minimizes false positives (Figure 2).
Software integration for code-free OCR deployment
Aurora Vision Studio provided a graphical environment where engineers could build and adjust workflows without coding. Its deep learning OCR model is pre-trained to recognize a variety of fonts, sizes, and orientations, even when the characters are distorted, worn out, or partially occluded. The ML610M and DFK 33UX178 captured text with consistent scale and sharpness, so Aurora required minimal adjustments to parameters such as text height and contrast.
Clean pixel data allowed DL_LocateText to accurately separate characters, while DL_ReadCharacters provided reliable recognition even in difficult industrial scenes. This functionality supports practical applications such as automated validation of product labels, real-time inspection of serialized parts, digitization of forms in harsh environments, and integration with robotic systems for sorting and inspection.
Operational results
The integration of high-resolution optics and deep learning software has resulted in measurable benefits. Although the ML610M lens is capable of higher resolution, the effective resolution of the system is set at 200 lp/mm by the DFK 33UX178 camera, which is more than enough to faithfully render the smallest text elements. This clean, high-contrast input allowed Aurora’s OCR engine to achieve reliable recognition, reduce false reads, and minimize the need for multiple captures.
Aurora Vision Studio required no code, allowing rapid deployment and allowing operators to make occasional adjustments without programming expertise. The system’s accurate imaging supports a variety of industrial OCR applications, including product serialization and traceability, label verification, electronics and PCB inspection, document digitization, and industrial automation.
As verified by application engineer Tommy Miner, the 17 pixels/mm density of the ML610M/DFK 33UX178 combination is exactly what Aurora’s deep learning OCR tools can perform with maximum reliability in real-world industrial conditions. Even at working distances closer than the lens’ nominal minimum of 0.5 m, the system provided crisp, high-resolution images at 9 inches (Figure 3).
The resulting vision system represents a robust OCR solution where optical precision and deep learning intelligence work together to support critical industrial applications, reduce errors, and improve operational efficiency.
See below for more information.
DFK 33UX178 – 33U Series Color Industrial Camera | Image Source
Theia ML610M True 4K Lens – Theia Technologies
Aurora Vision Studio™ | Machine Vision Software | Zebra
