Extended dynamic range
For industrial use, a wide dynamic range is for example an issue in the inspection of very bright scenes, where details of the process, as well as high richness of background details are required. A typical vision camera with 8 or 10 bits of dynamic range would only be able to see either the dark or bright areas with any detail. Using special support circuits or signal processing, cameras are able to acquire and output an increased dynamic range. Several technologies are available to achieve this:
With a linear sensor a sequence of images with different exposure times is acquired. These images are combined to an HDR image.
Some sensors can be used with an interleaved exposure mode. Odd and even lines are using different exposure times. These lines can be combined to an HDR image directly on the chip (e.g. with vertical Binning), by processing in an FPGA or on the host.
Multi-slope/piecewise linear response
This technology is available for some CMOS sensors. It is based on a reset pulse which is applied with different levels at selectable times during the total exposure time. Typically this reset can be done 1-2 times during one exposure period. The level of the reset pulse and the position relative to the exposure time defines the result of the HDR compression. The image below shows a typical sample of this mode:
Other technologies to create HDR images include:
- 2-chip cameras: 2 images with different exposure times.
- Dual-line line scan cameras: both lines are operated at different exposure times.
HDR images are either generated by software in the PC or using hardware processing (FPGA) afterwards.
The image on the right illustrates the effect when using this technique in a real situation. The whole scene is well exposed, even the filament of the lamp. Without using multi-slope exposure, either the lamp would be overexposed or the rest of the scene would be very dark and contain no details.