Home > A: Camera Basics > HDR (High Dynamic Range) > | History back Previous chapter Next chapter Print |
HDR Sensor Principles |
uEye Camera Manual Version 3.80
HDR Sensor Principles
Linear sensor with knee point characteristic
Some CMOS sensors provide the option of using a piecewise linear characteristic with knee points. With such a piecewise linear characteristic, the form of a gamma or logarithmic characteristic is roughly approximated.
|
Once a selected duration t1 (must be shorter than the exposure time) has elapsed, all pixels whose charge exceeds a selected value q1 are reset to the value q1. The exposure is continued normally. The following figure shows such a characteristic with one knee point.
Figure 53: uEye camera with HDR knee point function
In this case, the values are selected in such a way that all pixels whose current charge is more than 60% of the maximum value (saturation) are determined after 85% of the exposure time. These pixels are then reset to exactly this value of 60% of the maximum and exposed further. This means that pixels in bright areas of an image are saturated later, and the image has a higher dynamic range.
In the case of sensors with multiple knee points, multiple times tn and threshold values qn are defined. With this method, dynamic ranges of a calculated approximate value of 80 dB are achieved. The disadvantage here is that the knee points must be readjusted to each exposure situation and that overexposure of the sensor is still possible. In addition, this method is mainly suitable for monochrome sensors, as calculating the correct white balance using different knee points is very difficult with color sensors.
Logarithmic HDR sensor
Using the following formula (see also Dynamic Range ), it can be calculated that a linear 8-bit image with 256 brightness levels can no longer be imaged with a roughly 48 dB dynamic range:
To be able to display a dynamic range of 120 dB in an image with a linear characteristic, for example, a bit depth of 20 bits (1 million brightness levels) would be required accordingly. For this reason, the image data must already be available before digitization with a logarithmic characteristic that carries out dynamic compression in pixels. This is why a logarithmic characteristic is aimed for with HDR sensors.
Figure 54: Image capture with a logarithmic sensor