Inferensys

Glossary

Line Scan Camera

An image sensor that captures a single row of pixels at a time, building a continuous 2D image as the object moves, ideal for inspecting cylindrical parts or materials on a continuous web.
SRE continuously monitoring AI systems on multiple screens, real-time dashboards visible, dark mode NOC setup.
HIGH-SPEED IMAGING SENSOR

What is a Line Scan Camera?

A line scan camera is an image sensor that captures a single row of pixels at a time, building a continuous 2D image as the object or camera moves, ideal for inspecting cylindrical parts or materials on a continuous web.

A line scan camera is a specialized digital camera containing a single row of photosensitive pixels, rather than a 2D matrix. It constructs a complete image by assembling successive 1-pixel-high strips captured at high frequency as the target object moves perpendicularly across the sensor. This architecture eliminates frame overlap and geometric distortion on moving subjects, making it the standard for continuous web inspection of materials like textiles, paper, and steel, as well as for imaging rotating cylindrical components where a 2D area scan camera would produce skewed results.

The sensor's linear architecture provides extreme resolution in the scan direction, often exceeding 16,000 pixels per line, while the resolution in the motion direction is determined by the encoder-synchronized acquisition rate. Unlike area scan cameras, line scan systems require precise encoder synchronization to correlate each line capture with the object's physical position, ensuring a geometrically accurate image. This makes them indispensable for high-speed automated optical inspection (AOI) applications where surface defects must be detected on continuously moving products without stopping the production line.

ARCHITECTURE & APPLICATION

Key Features of Line Scan Cameras

Line scan cameras build a continuous 2D image one row at a time, making them the definitive choice for inspecting cylindrical parts, continuous webs, and high-speed conveyor systems where area scan cameras fall short.

01

Single-Line Sensor Architecture

Unlike area scan cameras that capture a full 2D frame at once, a line scan camera uses a single row of photosensitive pixels—typically 1,024 to 16,384 pixels wide. As the object or camera moves, the sensor captures sequential lines at high frequencies, assembling a theoretically infinite-length image. This architecture eliminates frame-boundary artifacts and provides geometric fidelity across the entire scan axis.

  • Pixel widths: 1K, 2K, 4K, 8K, 16K
  • Line rates: up to 200 kHz in modern CMOS sensors
  • Ideal for: webs, rolls, rotating cylinders
16,384 px
Max Sensor Width
200 kHz
Max Line Rate
03

Lighting Requirements & Geometric Constraints

The narrow, one-dimensional field of view demands high-intensity, focused illumination concentrated on a thin line. Standard diffuse lighting wastes energy outside the sensor's view. Common solutions include fiber-optic line lights, high-power LED bars, and laser line generators. The lighting geometry must be co-planar with the sensor to maximize photon collection at high line rates.

  • Bright-field: direct reflection for surface defects
  • Dark-field: angled illumination for scratches and texture
  • Laser profilometry: triangulation for 3D height maps
04

Unwrapping Cylindrical Objects

A defining application is inspecting the full surface of rotating cylindrical parts—bearings, pistons, rollers, and pharmaceutical vials. The part spins on a fixture while the camera captures lines, effectively 'unwrapping' the 360-degree surface into a flat 2D image. This reveals defects like scratches, pits, and inclusions that would be hidden or distorted in a single area-scan snapshot.

  • Rotation synchronized to line rate
  • Produces seamless panoramic surface maps
  • Enables 100% surface inspection at production speed
05

Time Delay Integration (TDI) for Low-Light Sensitivity

TDI line scan sensors contain multiple parallel rows of pixels that sequentially image the same line on the moving object. The charge from each row is summed in perfect synchronization with the object's motion, effectively multiplying the integration time without reducing speed. This provides dramatically higher signal-to-noise ratio in high-speed, low-light applications like web inspection of dark materials.

  • Common stages: 16, 32, 64, 128, 256 rows
  • SNR improvement: proportional to the number of TDI stages
  • Critical for: PCB inspection, semiconductor wafer scanning
256
Max TDI Stages
16×
Typical SNR Gain
06

Color & Multispectral Line Scan

Color line scan cameras use tri-linear sensors with three separate pixel rows, each filtered for red, green, and blue. Alternatively, prism-based cameras split the incoming light onto three discrete monochrome sensors for superior color registration. Beyond visible light, line scan cameras operate in UV, NIR, and SWIR wavelengths to reveal subsurface defects, material composition differences, and contaminants invisible to the human eye.

  • Tri-linear: cost-effective, requires spatial correction
  • Prism: precise co-site sampling, no interpolation
  • SWIR (900–1700 nm): sees through silicon, detects moisture
LINE SCAN CAMERA FAQ

Frequently Asked Questions

Clear, technically precise answers to the most common questions about line scan camera technology, its operational principles, and its role in modern industrial machine vision systems.

A line scan camera is a specialized image sensor that captures a single row of pixels at a time, building a continuous 2D image as the object or camera moves relative to the other. Unlike an area scan camera that exposes a full matrix of pixels simultaneously, a line scan sensor contains a single linear array of photodiodes. As the target object moves past the camera—typically on a conveyor belt, web, or rotating drum—the sensor reads out successive lines at high frequencies, often exceeding 50 kHz. These individual line acquisitions are then stitched together in software to form a seamless, theoretically infinite-length 2D image. This architecture makes line scan cameras ideal for inspecting continuous materials like paper, textiles, and metal strips, as well as cylindrical objects that are rotated during inspection. The key components include the linear CMOS or CCD sensor, a lens optimized for the narrow field of view, and a precision encoder that triggers line acquisition based on object movement to prevent geometric distortion.

HIGH-SPEED IMAGING

Industrial Applications of Line Scan Cameras

Line scan cameras build continuous 2D images one pixel row at a time, making them the definitive choice for inspecting cylindrical parts, continuous webs, and high-speed conveyor systems where area scan cameras fail.

01

Continuous Web Inspection

Line scan cameras are the standard for inspecting continuous materials like paper, textiles, and metal foil produced in rolls. As the web moves at high speed under the sensor, the camera captures a seamless, unbroken image without the frame overlap or gaps inherent to area scan cameras.

  • Printing Industry: Detects streaks, color registration errors, and missing print at speeds exceeding 100 meters per minute.
  • Non-Woven Textiles: Identifies holes, stains, and fiber density variations in real-time.
  • Battery Electrode Coating: Monitors the uniformity of lithium slurry coatings on copper and aluminum foils, where pinhole defects cause catastrophic cell failure.
100+ m/min
Typical Inspection Speed
02

360° Cylindrical Surface Inspection

Inspecting the entire surface of cylindrical components—such as bearings, pharmaceutical vials, or ammunition casings—requires unwrapping the object into a flat 2D image. A line scan camera synchronized with a rotating stage captures the full circumference in a single, distortion-free image.

  • Bearing Manufacturing: Detects surface pitting, scratches, and grinding burns on roller elements and raceways.
  • Pharmaceutical Packaging: Inspects the sidewalls of glass vials and syringes for cracks, cosmetic defects, and particulate contamination.
  • Ammunition & Aerospace: Verifies the integrity of casings and fasteners, where surface flaws can lead to catastrophic failure under extreme pressure.
360°
Full Surface Coverage
03

High-Resolution Flat Panel Inspection

The manufacture of flat panel displays and printed circuit boards demands pixel-level defect detection across meter-wide substrates. Line scan cameras with high-density sensors and telecentric lenses provide the necessary resolution and geometric accuracy.

  • OLED/LCD Panels: Identifies dead pixels, mura (non-uniformity), and color filter defects on substrates exceeding Gen 10.5 dimensions.
  • PCB & Semiconductor Wafers: Detects micro-cracks, solder bridging, and trace width violations with micron-level precision.
  • Solar Cell Manufacturing: Inspects photovoltaic wafers for micro-cracks and finger interruptions that reduce energy conversion efficiency.
< 5 µm
Achievable Pixel Resolution
04

Free-Fall Sorting & Food Grading

In free-fall optical sorting, products cascade past a line scan camera at high velocity. The camera captures a snapshot of each object's entire surface as it passes, and downstream air jets eject defective items based on real-time classification.

  • Grain & Seed Sorting: Removes discolored, moldy, or foreign-material kernels from bulk food streams at throughputs measured in tons per hour.
  • Recycling & Waste Management: Separates plastics by polymer type using near-infrared (NIR) hyperspectral line scan cameras.
  • Frozen Food Processing: Detects foreign material and color defects in individually quick-frozen vegetables and fruits.
Tons/Hour
Sorting Throughput
05

Rail & Infrastructure Monitoring

Line scan cameras mounted on track inspection vehicles capture continuous, high-resolution images of rail surfaces, overhead catenary wires, and tunnel linings at mainline speeds. The resulting image strips are analyzed for fatigue cracks, wear patterns, and geometric deviations.

  • Rail Surface Inspection: Detects rolling contact fatigue cracks, corrugation, and squats that precede rail breaks.
  • Overhead Wire Monitoring: Measures the stagger and wear of contact wires to prevent pantograph entanglement.
  • Tunnel & Bridge Surveying: Creates high-resolution surface maps to detect concrete spalling and water ingress over time.
300+ km/h
Inspection Speed Capability
06

Synchronization & Encoder Integration

Unlike area scan cameras that capture discrete frames, line scan cameras require precise synchronization with the motion of the object. A rotary encoder mounted on the conveyor or web drive outputs pulses that trigger each line acquisition, ensuring the aspect ratio remains correct regardless of speed variations.

  • Quadrature Encoders: Provide direction and speed data, allowing the camera to maintain spatial accuracy during acceleration and deceleration.
  • Line Rate Calculation: The required line rate (in kHz) is determined by dividing the object speed (mm/s) by the desired pixel resolution (mm/pixel).
  • Image Distortion: Without encoder synchronization, speed fluctuations cause compression or stretching artifacts, rendering dimensional measurements invalid.
200+ kHz
Maximum Line Rate
SENSOR ARCHITECTURE COMPARISON

Line Scan vs. Area Scan Cameras

Technical comparison of 1D line scan sensors and 2D area scan sensors for automated optical inspection applications.

FeatureLine Scan CameraArea Scan Camera

Sensor Geometry

Single row of pixels (1D)

2D matrix of pixels

Image Acquisition

Builds image line-by-line with motion

Captures entire frame simultaneously

Resolution (Cross-Web)

Up to 16K+ pixels per line

Typically 5-45 megapixels

Ideal Subject

Continuous webs, cylinders, rotating parts

Stationary or discrete objects

Lighting Requirements

Single focused line of illumination

Full-field uniform illumination

Frame Rate

Up to 200+ kHz line rate

Typically 30-340 fps

Distortion on Curved Surfaces

Minimal; each line is tangent

Significant perspective distortion

Integration Complexity

Requires encoder synchronization

Simple trigger-based capture

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.