We have been working on an autoexposure (AE) and autowhite balance (AWB) support for Davinci™ DM36x family of processors using the hardware support offered by the image pipe (IPIPE). Some sensors (like the MP9P031 5MP sensor) lack AE/AWB on chip, so you need to implement it by yourself. TI's Davinci and OMAP™ SoCs families have support for hardware accelerated AE/AWB through the IPIPE, and the PSP kernels offer char drivers support for the kernel, however programming them and implementing the algorithms efficiently is not a trivial task. Melissa, one student from TEC (and now a RidgeRun employee), implemented several AE/AWB algorithms that take advantage of the IPIPE hardware using a fixed point arithmetic that is ideal for low-power DM36x ARMv5 processors. RidgeRun is commercializing the AE/AWB library for TI's customers who need AE/AWB support. The library is plain C code and can be integrated with any Linux application. Doxygen documentation is available. Melissa also worked along with some other RidgeRun employees to create an open source project called Image Pipe Daemon (ipiped). It provides a daemon with a DBus API to control the behavior of the camera subsystem from other applications (sample cli tool provided). ipiped is open source under BSD license and provides a nice architecture to interact with the image pipe hardware (and is designed to be modular and easily adapted to other image pipe architectures like the ones on the OMAP family). Furthermore ipiped integrates nicely with RidgeRun's AE/AWB library . Be sure to contact us if you need more information.