Basic principles of image sensors

Color

Capturing color images is sometimes necessary. Two techniques are mainly used to digitize color: either the resolution of only one sensor is decimated by a color filter network (Bayer filter, 3 colors, Fig. 35, or more, Fig. 36), or the light is dispersed into several components by a system of prisms in which each one produces the image of the object's color component on a detector dedicated to this color (Fig. 37).



   

     Figure 35: Bayer filter in three colors
Figure 35: Bayer filter in three colors [zoom...]Info

The final image is then reconstructed using algorithms that recombine the components and treat color sensitivity, gain and spectral decomposition of the light signal. These algorithms guarantee the «depixelation» of figure 38.



   

     Figure 36: Bayer filter in four colors on CMYG real sensor (Photo EMA)
Figure 36: Bayer filter in four colors on CMYG real sensor (Photo EMA) [zoom...]Info


   

    Figure 37: Color separation using prisms (source Dalsa)
Figure 37: Color separation using prisms (source Dalsa) [zoom...]Info


   

    Figure 38: Reconstruction of a color image
Figure 38: Reconstruction of a color image [zoom...]Info

The last possibility consists in piling the light sensitive diodes and filtering by selecting the wavelength on the sensor depth (Fig. 39).



   

     Figure 39: Principle of the Foveon sensor
Figure 39: Principle of the Foveon sensor [zoom...]Info
AccueilNouvelle pageInformations sur le cours (ouvrir dans une nouvelle fenêtre)Choosing the sensorSet-up