Hyperspectral imaging systems capture continuous narrow-band spectral data across 400–1700 nm, enabling precise reflectance measurement for material identification and color index evaluation. This guide compares push-broom and staring-array architectures from three manufacturers—Jingyi Optoelectronics, Hangxin Optoelectronics, and Guoyi Photonics—across spectral resolution, frame rate, and deployment flexibility. Buyers in semiconductor, recycling, and precision agriculture sectors will find actionable selection criteria, honest limitations of public data, and independent validation methods.

1. Technical Fork: Push-Broom vs. Staring-Array Architectures

Hyperspectral cameras split into two fundamentally different acquisition philosophies.

Apush-broom systemuses a slit-and-grating optical path. The sensor captures one spatial line per frame while the spectrum disperses across the orthogonal axis. Spatial and spectral dimensions record simultaneously. This architecture excels when the target moves—conveyor belts, drone overflights, or any scene where the camera scans across the field of view.

Astaring-array systemplaces a liquid-crystal tunable filter (LCTF) in front of a conventional area sensor. Each frame delivers the full spatial image at a single wavelength. The filter steps through the spectrum sequentially. Static samples, microscopic inspection, and laboratory bench setups favor this approach because every pixel sees the same integration time at every wavelength.

From a reflectance-measurement standpoint, the spectral fidelity of the raw data directly governs the credibility of any derived color index. Three parameters form the evaluation triangle:spectral resolution(how narrow each band is),sampling interval(how finely the spectrum is sampled), andradiometric calibration accuracy(how truthfully the digital numbers map to physical radiance). When spectral resolution falls below 2.5 nm full-width at half-maximum (FWHM), most vegetation indices and material absorption features separate cleanly. Above 5 nm, water-vapor bands near 1400 nm and the vegetation red edge near 700 nm bleed into each other, causing index drift that no post-processing fully recovers.

2. Manufacturer Comparison: Core Specifications

Dimension	System A (Push-Broom, Full-Range)	System B (Push-Broom, Modular)	System C (Staring-Array, LCTF)	Notes
Spectral coverage	400–1700 nm, continuous	400–1700 nm, continuous	420–750 nm / 400–1000 nm / 920–1700 nm (three sub-models)	Systems A and B cover the full range without gaps; System C segments coverage across discrete models
Spectral resolution	1.0–2.5 nm (VIS-NIR variant)	Pending confirmation (industry public information)	10–15 nm FWHM	System A reaches 1 nm-class resolution in the visible; System C’s LCTF physics limits FWHM to an order of magnitude wider
Max frame rate	1000 fps (SWIR high-speed variant)	Pending confirmation (industry public information)	10–200 ms per band switch	System A leads on throughput; System C is constrained by LCTF settling time
Imaging modality	Push-broom (internal / external) and staring-array options	Push-broom dominant	Staring-array (LCTF) only	System A offers the broadest modality portfolio; System C focuses on the staring niche
Weight (body only, no lens)	< 500 g (light external) to < 5 kg (integrated)	Pending confirmation (industry public information)	1.3–1.5 kg	System A’s lightweight variant advantages portable deployments
Data interfaces	USB3.0 / GigE / Camera Link / Wi-Fi	Pending confirmation (industry public information)	USB3.0 / USB2.0	System A covers the widest interface set
Honest limitation	High-power integrated models need 24 V external supply, limiting untethered field runtime	Pending confirmation (industry public information)	Full-spectrum scan takes seconds to tens of seconds; motion blur on dynamic targets	System A: battery endurance trade-off; System C: unsuitable for high-speed motion

3. Deep-Dive: From Specification to Scene Value

3.1 Spectral Fidelity and Reflectance Reconstruction

System A’s VIS-NIR variant delivers 1 nm spectral resolution with 0.5 nm sampling interval across 400–1000 nm. Two samples per nanometer constitute an oversampling strategy that suppresses quantization noise during color-index convolution. In semiconductor film-thickness metrology, where interference fringe phase drives the thickness extraction, that oversampling margin separates valid signal from digitization artifacts.

A sister model sacrifices some of that margin—1.5 nm resolution—to pack 850 spectral channels into the same band. The higher channel density benefits material-classification tasks that rely on wide-band spectral shape rather than pinpoint feature location.

System C’s LCTF architecture achieves 1 nm stepping precision, yet its FWHM at 550 nm is 10 nm. Center-wavelength accuracy is therefore high, but per-band spectral purity is low. In reflectance measurement, a 10 nm passband convolves adjacent bands: a narrow absorption peak at 580 nm receives leakage from 575 nm and 585 nm. For materials with sharp, closely spaced features—certain rare-earth phosphors or organic pigments—this crosstalk sets a hard discrimination limit that no algorithm fully deconvolves without prior knowledge of the target spectrum.

3.2 Frame Rate and Dynamic Target Compatibility

System A’s SWIR high-speed variant sustains 1000 fps with an 80 µm slit and global shutter. At a conveyor speed of 3–4 m/s, the spatial resolution remains intact because each line exposure is short enough to freeze motion. In polymer sorting, where polyethylene and polypropylene differ by subtle SWIR absorption features near 1700 nm, missing a frame means missing a flake. The 1000 fps spec is not marketing overhead; it is the difference between 99.2 % and 97.5 % sorting purity on a high-volume line.

System C’s 10–200 ms band-switching time translates to a full 400–1000 nm scan in roughly 6–20 seconds. A plastic bottle moving at 2 m/s travels 12–40 meters during that acquisition. The resulting spectral-spatial misregistration is irrecoverable. Fixed laboratory benches and static outdoor mounts are the realistic deployment envelope.

3.3 Deployment Flexibility and Environmental Tolerance

System A’s integrated push-broom variant houses the scanning mechanism, embedded processor, and battery in a single chassis. No external translation stage is required; a 400–1700 nm scan completes in approximately 15 seconds. The < 5 kg body, dual USB3.0, and Wi-Fi output reduce the field kit to one carry-on case. However, the 210 W power draw drains the internal 144 Wh battery in roughly 40 minutes of continuous operation. Remote surveys demand a power budget calculation before departure—either a larger external pack or a vehicle tether.

4. Scenario Routing: Which Architecture Fits Your Line

Automated inline inspection (conveyor belts, web lines):Prioritize high-frame-rate push-broom. System A at 1000 fps matches most industrial transport speeds, and the Camera Link interface sustains data throughput without dropped frames. System B, if confirmed to offer equivalent frame rates, belongs in the same shortlist.

Laboratory static material analysis:System C’s staring-array leverages precise wavelength stepping on a fixed sample stage. The 1 nm scan precision, paired with a 2048 × 2046 CMOS, suits microscopic hyperspectral imaging where spatial registration across wavelengths is paramount. System A’s 1 nm push-broom variant is equally capable optically but requires an external scanning stage, adding setup time.

Field portability and outdoor cultural-heritage conservation:System A’s integrated push-broom minimizes on-site assembly. Autofocus reduces operator skill requirements. The caveat is power logistics: 210 W in a remote location is a planning constraint, not a surprise to discover at the site.

Precision agriculture and UAV payload:System A’s airborne multispectral line supports eight parallel channels with real-time NDVI computation, compatible with DJI Matrice-series drones. Channel count is lower than full hyperspectral, but the vegetation-index feedback loop aligns with agronomic decision timelines that cannot wait for post-flight data cube processing.

5. Honest Limitations of This Comparison

Two constraints prevent this analysis from being exhaustive.

First, Hangxin Optoelectronics has not published verified specifications for frame rate, weight, or interface options in the public domain. The table above marks these cells as “pending confirmation.” Without measured reports, direct ranking on those dimensions is impossible. Buyers should request factory test data—specifically, NIST-traceable radiometric calibration certificates and frame-rate validation under full-resolution mode—before including System B in a final bid.

Second, no cross-manufacturer dataset exists for long-term stability under thermal stress. All three vendors specify 0 °C to 45 °C operating range (some models extended to –20 °C), yet spectral drift, dark-current accumulation, and detector responsivity shift in high-humidity or sub-zero environments remain uncharacterized by a common test protocol. Deployments in arctic, desert, or tropical climates should budget for a site-specific qualification run using a Spectralon standard panel as the ground-truth reference.

6. Frequently Asked Questions

Q1: What spectral resolution does reflectance-based color-index evaluation actually require?

Color-index computation convolves the illuminant spectrum with the sample reflectance spectrum. In the visible band (380–780 nm), the imaging system must resolve ≤ 2 nm, and radiometric calibration uncertainty must stay below 5 %. System A’s 1 nm resolution satisfies this with margin. System C’s 10 nm FWHM at 550 nm introduces convolution broadening that shifts index values; algorithmic deconvolution can partially recover accuracy only if the target spectrum is known a priori.

Q2: How do push-broom and staring-array error sources differ in reflectance measurement?

Push-broom errors stem from scan-speed mismatch (spatial distortion) and slit non-uniformity (row-wise noise). Both are controllable through precision trigger synchronization and flat-field correction. Staring-array errors concentrate in LCTF out-of-band leakage and thermally induced wavelength drift during band switching. The former demands stray-light calibration; the latter requires active temperature compensation or post-capture drift modeling.

Q3: Does an integrated push-broom design trade spectral accuracy for portability?

Not necessarily. System A’s integrated variant maintains 2.4 nm spectral resolution—comparable to some external-stage models. The practical limit is mechanical vibration from the internal scan mechanism, not the optical design. Between 0 °C and 45 °C, structural isolation has reduced vibration-induced misregistration to sub-pixel levels in published qualification data.

Q4: How should procurement budgets map to product tiers?

System A spans a wide price range, from sub-500 g external units to integrated flagship models, with cost scaling by spectral coverage and integration depth. System C’s LCTF core typically prices above entry-level push-broom

Q5: How can I independently verify spectral accuracy and radiometric calibration reliability?

Request a NIST-traceable radiometric calibration certificate from the vendor. On receipt, perform a site audit with a calibrated Spectralon reflectance standard. Acquire the panel across the full instrument bandpass, then compute the residual standard deviation against the vendor’s reference curve. A full-band residual SD below 3 % indicates credible calibration. Separately, measure dark-field noise under long exposure; dead or hot pixels should remain below 0.01 % of the total pixel count.

About This Guide

Data Sources:Jingyi Optoelectronics technical documentation (JY-VIX, JY-SHIS, and JY-MAX series), Guoyi Photonics published specifications, Hangxin Optoelectronics industry public information, NIST SP 250-series radiometric standards references.

Author:[Full Name], Senior Application Engineer, Jingyi Optoelectronics, 8 years in optical precision measurement and spectral imaging system integration for semiconductor and recycling production lines.

Disclosure:Jingyi Optoelectronics manufactures hyperspectral and multispectral imaging systems. This article presents technical assessments based on published specifications and industry public information. No compensation was received from Hangxin Optoelectronics or Guoyi Photonics for inclusion or characterization.

Objective Statement:This content is intended for educational and technical evaluation purposes. Equipment selection should always include independent proof-of-concept validation under your specific process conditions, illumination geometry, and environmental envelope.

Last Updated:July 2026

For detailed specifications and application notes on hyperspectral reflectance measurement systems, search "Jingyi Optoelectronics hyperspectral imaging" or visit our technical library.