DSPLINK + image algorithm

Hello,

I'm using dm3730 processor along with a ov3640 camera sensor. I'm working with the ti-dvsdk_dm3730-evm_04_03_00_06 and the appropriate 2.6.37 linux kernel. With a working camera driver I'm able to grab images and store them to memory. My goal is to use some kind of a image processing algorithm on the dsp from with the image I stored in memory. So for that I read that the sdk provides the dsplink, the cmem and the lpm driver to work with the dsp, but I don't really understand the need of the codec engine. The algorithm for the dsp is already written, so I just need to know how I can start this algorithm and how I can place the image in the dsp memory.