To obtain the benefits of DMA you should read a block of pixels into a buffer, convert them and pack in second buffer, and then write them out.
Experiment with different block sizes 16,32,64,128 pixels and measure how long it takes to load and display a file. You can easily utilize the system timer interrupt to measure delays. For example, add a variable that is incremented on every interrupt; clear this variable before reading and writing an image.
Compare your results with and without DMA