Line Scan Cameras

The difference between an area camera and a line scan camera is not resolution or speed, it is how the image comes into being. An area camera captures a two dimensional frame at an instant. A line scan camera captures one row at a time, repeating as the product moves, and the image is assembled from those rows. Everything distinctive about the technology follows from this: it needs something to be moving, it needs to know how fast that thing is moving, and the spatial relationships in the finished image are a record of the motion as much as of the product.

Where it is the only answer

• Continuous material. Film, foil, paper, fabric, wire, cable, extruded profile, sheet metal from a coil. The material does not end, so there is no frame to take. An area camera could only sample it in pieces, with seams and gaps between them.
• Very long parts. A part several metres long, measured for defects along its whole length, would need either a very large sensor or a mosaic of area cameras. A line scan camera covers it with one sensor and one lens as it passes.
• Cylindrical surfaces. A rotating part passes a line scan camera and unrolls into a flat image — the surface of a shaft, a can, a bottle, a bearing race. Try that with an area camera and you get a fisheye view of one side and nothing of the rest.
• Very high resolution in one direction. Where the requirement is a great many pixels across a narrow strip, a line scan sensor gives more of them for the money than a comparable area sensor.

The counterweight is that everything must line up. The product has to move at a speed the system knows, the lighting has to be intense and continuous enough to freeze a moving row, and the line rate has to be locked to the motion. A line scan station that is set up well is excellent; one where the encoder signal is noisy or the speed varies is worse than an area camera, because its failures are geometric and therefore hard to see.

Speed and line rate have to agree

The relationship is simple and unforgiving. If the product moves at a speed of V millimetres per second and you want a resolution of R millimetres per line across the direction of motion, the line rate must be V divided by R lines per second. Get it wrong and the image is distorted: too few lines and the product looks stretched along its length, too many and it looks compressed. Either way, a circular feature becomes an ellipse and every measurement along the motion direction is scaled by the same wrong factor.

This table hides the real constraint. Each line is an exposure, and a short exposure at a high line rate means very little light per line, so the illumination has to be correspondingly intense. In practice the light source, not the sensor, sets the speed limit on a line scan station. This is why line scan setups so often use high intensity bars or laser illumination rather than the diffuse panels an area camera would use.

Practical checks before committing

Before choosing a line scan camera, confirm three things on the real line. That the product motion is consistent enough over the length being inspected — measure it, do not assume. That the lighting can deliver the intensity the line rate requires at the exposure time the speed allows, which usually means a trial with the actual source. And that the encoder can be coupled to the material rather than to something upstream of the slip, because that coupling determines whether the geometry is trustworthy. If any of the three is uncertain, an area camera with a mosaic, or a different inspection position, is often the cheaper answer.