Column noise

Lider is based on a mathematical technique called regularization, which is commonly used to solve many image restoration problems (such as deblur, denoise, etc.)

Using this type of technique, it is possible to perform column noise reduction on infrared images given by bolometer sensors.

Origin of problem and usual solution applied

These columns come from a gain drift which is different for each column of the infrared sensor. The usual method used to cope with this gain drifting problem relies in closing the camera aperture with a mechanical obturator and performing a new sensor calibration every now and then.

Numerical Solution

It is based on a restoration technique, involving mathematical regularization, which filters out column noise. In order to do so, one can apply a global or a local approach.

Results obtained through a global approach

The following image shows column noise on it as seen through an infrared camera. By rolling your mouse over this picture, you will see the change obtained through digital processing of the entire image. We denote a low frequency wavelike pattern superimposed on the image as a result of this processing. However, we can get rid of this pattern in real-time by changing continuously the regularization parameter thanks to screen or eye persistence.


© Copyright2005-2009 - LIGHTNICS /// 33(0) ///