I have an image of 16u values. These values are indexes into a map of 8u values, so the map is a lookup table for converting the u16 values to 8u values. Is there a function for this in IPP?
Also, another question. Say I have a one channel image of 32f values and I want to do general arithmetics on it (add,mul,sqrt...), would it be faster to use the image functions or the signal processing functions? There is no ROI involved in the calculations.
regarding your first question I think you may review of ippiLUTPalette functions.
I would recommend you to try Intel DMIP (Deferred Mode Image Processing)library for implementation image arithmetic tasks. It will build optimial processing pipeline for you and will alsoparallelize the task.
In ippi the data are in row major order and I use column major in my data since I do alot of column by column. My data is more or less a collection of vectors. Does there exist a similar function to LUTPalette in ipps?