High dynamic range (HDR) images can be composed by capturing a series of pictures of the same scene but of different exposure time. The more pictures used to create the HDR image, the more details that are likely to be preserved in the final picture. However, taking more pictures requires more exposure time, thus being volatile to camera shake or moving objects in the scene. On the other hand, smarphone cameras are more and more widely used in our everyday life. By integrating the HDR functions on the smartphone, images can be captured, rendered and customized in a more user-friendly way. The open-source nature of Nokia N900 makes it possible to have control over the hardware of the camera.

Images

		Captured images: Ten images captured at different expsoure time of the same scene in a neighbourhood in a late afternoon.
		Response function of Nokia N900: The LUTs for the red, green and blue channels. The three curves align well with each other. For more details, see Debevec 1997.
		Weighting function: More weight is given to digital counts in the middle range. Darker pixels receive less weight given smaller signal-to-noise ratio. On the other hand, pixels that are too bright are more likely to over-saturate. The hat weighting function in the paper (Debevec 1997) does not work very well in this scene.
		The HDR image: The HDR image rendered given the captured images.

Slides

How to set up FCam on Nokia N900

Related Project

Recovering High Dynamic Range Radiance Maps from Photographs