Line-Scan Hyperspectral Camera: High-Precision Spectral Capture for Industry and Research

Line-scan (push-broom) hyperspectral imaging is the most mature technology route in the field of hyperspectral imaging. It relies on a grating to obtain the continuous spectral information of each row of pixels of a target. Thanks to its excellent signal-to-noise ratio, spectral resolution, and imaging stability, it has long been the mainstream solution for precise laboratory analysis and online inspection in industrial production lines. While portable snapshot hyperspectral devices have quickly expanded on-site detection scenarios due to their compact size, snapshot solutions still have limitations in terms of data accuracy for large field-of-view sample inspection, high-speed continuous material sorting, and precise spectral acquisition in low-light environments. Line-scan hyperspectral equipment can balance spectral integrity and imaging efficiency, making it suitable for detailed analysis of static samples in the lab as well as integration into automated production lines or airborne platforms for continuous spectral data collection of moving targets.

I. Line-scan Hyperspectral Sensors: Core Technology Route

The core principle of line-scan (push-broom) hyperspectral imaging is to capture the complete continuous spectrum of one spatial row of the target at a time using a slit and grating system, and then scan row by row through the relative movement between the camera and the target to generate a hyperspectral data cube with 2D spatial information + 1D spectral information. The FS-1X color spectrum series all adopt a transmissive grating architecture. The incoming light is collimated through a slit, then dispersed by the transmissive grating, and finally projected onto a line detector for photoelectric conversion.

The transmissive grating optical path has strong stray light suppression (stray light can be less than 0.5%), minimal spectral distortion, and high light throughput, making it an ideal choice for industrial online inspection and scientific research scenarios. According to different detection bands, detectors are divided into CMOS and InGaAs types: visible to near-infrared band (400–1000nm) uses a CMOS detector; shortwave infrared band (900–1700nm, 1400–2500nm) uses an InGaAs infrared detector to match the photon response characteristics of different bands. 

There is a fundamental difference between line-scan imaging and snapshot area imaging: a snapshot camera captures complete 3D data in a single exposure, while line-scan imaging requires relative motion to complete full imaging. The motion synchronization accuracy directly affects image geometric distortion and spectral registration quality. With device iterations, the FS-1X series has achieved a lightweight design, with some shortwave infrared models weighing as little as 625g, retaining the high-precision advantages of transmissive gratings while meeting installation constraints for production line integration or platform deployment.

Main FS-1X model band configurations:

• FS-10: 400–700nm (visible light)
• FS-12/FS-13: 400–1000nm (visible to near-infrared)
• FS-14: 400–1700nm (visible + shortwave infrared)
• FS-15/FS-17: 900–1700nm (shortwave infrared)
• FS-1A: 1400–2500nm (longwave shortwave infrared)

II. Performance Trade-offs of Line-Scan Architecture: Key Metrics to Consider for Selection

The performance parameters of line-scan hyperspectral systems balance each other out, and there is no model that excels in every parameter. Selection needs to be based on actual working conditions. Based on the FS-1X series hardware characteristics, the core evaluation metrics are as follows:

Spectral Range:

• 400–700 nm (visible light), suitable for color and texture appearance inspection of printed materials and textiles;
• 400–1000 nm (visible to near-infrared), commonly used for ore identification, liquid recognition, fruit and vegetable sorting;
• 900–1700 nm (shortwave infrared) can identify medicinal herbs, polymer materials, and ore components;
• 1400–2500 nm (longwave infrared) is used for analyzing special minerals and polymer materials.
• The wider the spectral range, the higher the difficulty in optical design and the overall system cost.

Spectral Resolution (FWHM): This indicates the ability to distinguish adjacent spectral absorption features.

• FS-10/12/13 visible to near-infrared models can reach a spectral resolution of 2.5 nm;
• FS-17 shortwave infrared model has a resolution of 6 nm;
• FS-14 cross-band model 13 nm;
• FS-1A longwave model 12 nm.
  
• Fine material identification and quantitative composition analysis require higher spectral resolution, but increasing resolution narrows the slit, which negatively affects the signal-to-noise ratio.

Signal-to-Noise Ratio (SNR): Most FS1X models can achieve a peak SNR of 600:1, while longwave FS-1A is 300:1. SNR is influenced by slit width, light source brightness, and line frequency (exposure time); when production lines run at high speed, line frequency increases and exposure time per line shortens, causing the SNR to drop. In industrial scenarios, a balance between acquisition speed and data quality is required.

Line Frequency / Imaging Frame Rate and ROI Multi-Region Selection: Imaging line frequency directly relates to the material conveyor speed allowed by the production line.

The FS-1X series supports ROI band gating, and FS-13, FS-17, FS-14 support multi-region band selection, outputting only the required spectral bands. This can greatly increase acquisition frame rates; for example, FS-13 full-spectrum is 128 Hz, but after ROI selection, it can reach up to 3300 Hz, meeting high-speed production line needs.

Spatial Sampling, Interfaces, and Mechanical Installation

The number of spatial pixels determines the lateral field of view for scanning. All models use standard C-Mount lens interfaces, allowing flexible selection of different focal length lenses. Output interfaces include USB3.0/3.1 and Gigabit Ethernet, compatible with different host computer transmission needs. Overall system weight ranges from 625 g to 4.4 kg, so installation bracket load capacity and device space must be considered.

Motion Synchronous Control

Line-scan imaging heavily depends on synchronous triggering. Industrial production lines need to connect with encoders to trigger according to conveyor speed; lab electric stages and airborne platforms also require position and attitude signal synchronization. Mis-synchronization can lead to image stretching and spectral registration errors.

III. Main Application Scenarios of Line-Scan Hyperspectral Sensors

3.1 Industrial Assembly Line Sorting and Machine Vision

Line-scan hyperspectral sensors are suitable for detecting continuously moving materials. The FS-1X series is widely used for inspecting printing defects, checking textile colors and fabrics, detecting screen defects, sorting fruits and vegetables online, and classifying plastic materials in solid waste. By combining spectral and imaging capabilities, they can detect component differences that regular RGB cameras can't see, enabling foreign object removal and quality grading. Single-pixel color measurement reproducibility is dE*ab < 0.1, meeting the strict requirements of industrial color inspection.

3.2 Laboratory Sample Composition and Material Identification

With an electric displacement scanning stage, they can classify ores, analyze drug components, and identify powder and liquid samples. For example, the FS-13 can capture ore sample spectra and identify liquid components, distinguishing water from alcohols; FS-15 can cluster and identify ores and differentiate spectra of Gastrodia elata powder at different concentrations, achieving non-destructive qualitative identification based on unique spectral fingerprints.

3.3 Agriculture and Agricultural Product Quality Detection

Used for checking internal quality of fruits and vegetables, authenticating medicinal materials and their concentration, and detecting mold in produce. They can be set up as benchtop detection platforms for small sample experiments or integrated into sorting equipment for large-scale online screening.

3.4 Geological and Material Analysis

Short-wave infrared models FS-15, FS-17, and FS-1A are designed for mineral and rock samples, identifying spectral features of different ores and used for ore sorting and rapid mineral component screening.

IV. Market Trends and Key Points for Avoiding Pitfalls in Selection

Industry Trends

1. Lightweight line scan models continue to evolve. For example, the FS-17 short-wave infrared model weighs only 625g, significantly reducing installation load. Line scan devices are gradually moving beyond traditional large equipment roles and are being widely integrated into automation equipment.

2. Multi-region ROI band selection has become mainstream, outputting only the business-relevant bands. This improves acquisition frame rates without sacrificing core spectral information and suits high-speed production lines.

3. Standardized C-Mount interfaces and open SDKs allow flexible lens and light source selection, reducing the workload of custom development.

Key Points for Avoiding Pitfalls in Selection

• Don’t blindly pursue extreme spectral resolution: Higher resolution can reduce light efficiency. If on-site lighting is limited, this will directly lower the actual signal-to-noise ratio, and real-world data may worsen.
  
• Confirm synchronous triggering solutions: Line scan image quality heavily depends on motion synchronization. Without a reliable triggering mechanism, even high-end hardware won’t produce qualified spectral data.
  
• Distinguish between hyperspectral and multispectral: FS-1X line scan hyperspectral outputs hundreds to over a thousand continuous spectral channels. Multispectral has only a few discrete bands, with a significant difference in material recognition capability. Don’t confuse them when purchasing.
• Verify ROI capability and interfaces: High-speed production lines should prioritize models supporting multi-region ROI. Confirm output interfaces, SDK development support, and assess secondary development workload.
• Pay attention to detector differences: Visible to near-infrared uses CMOS, while short-wave infrared uses InGaAs. Different detectors have obvious differences in signal-to-noise ratio, weight, and cost. Selection should consider the spectral absorption characteristics of the target material.

V. Summary

Push-scanning line scan hyperspectral systems, relying on transmissive grating dispersion technology, provide low stray light, high signal-to-noise ratio, excellent spectral resolution, and reproducibility. They play an irreplaceable role in industrial online inspection, laboratory material analysis, agricultural product screening, and geological material identification. However, this architecture is constrained by relative motion and requires strict synchronous control and platform stability.

The FS-1X represents a new generation of lightweight line scan hyperspectral models, covering the full spectral range from visible light to 1400–2500nm, balancing performance and size constraints for different scenarios. Selection should not focus solely on maximizing one parameter but consider spectral range, spectral resolution, line frequency, signal-to-noise ratio, and installation conditions. A complete project requires coordinated hardware, light source, synchronous control system, spectral software, and algorithms. Conducting sample pre-tests early in the project is the most effective way to mitigate risks.