Line scan is the right architecture whenever the material is continuous and the defect must be found before it is wound up or cut. Instead of a rectangular sensor, a line scan camera has one or a few very long rows of pixels, and it builds the image one line at a time as the material passes. That single design decision changes the maths, the lighting and the mechanics.
Line rate is a speed calculation, not a camera spec
The camera's line rate is only meaningful in combination with the material speed and the resolution you have chosen. Two constraints have to be met simultaneously:
Resolution constraint. To detect a 0.2 mm defect you need at least two, preferably three, pixels across it, so 0.07 mm per pixel. A 2048-pixel line at 0.07 mm per pixel covers 143 mm of material width.
Rate constraint. At 0.07 mm per pixel and a material speed of 1 m/s, the camera must expose and output 1000 divided by 0.07 = 14,286 lines per second. That number, not the pixel count, decides whether a given camera can do the job.
Line scan cameras such as the MV-CL022-91GM and the higher-rate MV-CL042-91GC are specified at line rates in the tens of thousands per second precisely because this is the governing constraint. If your calculation lands above the camera's maximum, the only options are a coarser pixel size, more light, or splitting the field of view across two cameras.
Exposure has to be short, so light has to be strong
At 14,000 lines per second the time available per line is about 70 microseconds. Exposing for the full 70 us would smear the image badly at 1 m/s. Practical line scan exposure times are a small fraction of the line period, which means the instantaneous illuminance must be very high - typically a high-intensity LED bar or line light running continuously, or a strobe synchronised to the line clock.
Line lights are not interchangeable with area lights. The illumination must be a narrow, uniform bar aligned with the camera's field of view, and uniformity along that bar directly becomes brightness variation across the image. This is why line scan projects spend a disproportionate share of their budget on illumination.
Encoders, and why they are not optional
Line scan cameras must be triggered by an encoder locked to the material, not by a free-running internal clock. If the camera lines are timed independently of the web, any speed variation stretches or compresses the image, and the same physical feature appears at different sizes on different days. Encoder-locked triggering guarantees constant millimetres per line regardless of line speed, which is what makes measurements from a line scan image meaningful at all.
The encoder must be mechanically coupled to the material, not to the drive motor - a motor encoder measures commanded motion, and belt slip between motor and web is exactly the error you are trying to avoid. A measuring wheel riding on the material is the usual solution.
Optics and mechanics for line scan
Line scan sensors are long and narrow, which strains ordinary lens designs: a lens must resolve across a 2048 or 4096 pixel line without falling off in sharpness at the ends. High-resolution fixed focal lenses such as the MVL-KF5028M-12MP handle moderate line lengths well; longer lines on wide webs generally need dedicated large-format lenses.
Mechanically, the camera must be perpendicular to the direction of travel within a fraction of a degree, or the image will be sheared. And because line scan systems accumulate the image over time, any vibration becomes a wavy line rather than a blurred patch - rigidity matters more than it does for area scan.
Line scan is not harder than area scan because of the camera. It is harder because it demands a complete, coherent system: encoder, line light, rigid mounting and a line rate calculation that actually closes. Do that arithmetic before choosing a model and the rest of the project is predictable.