If both images are the same depth, size etc, use 'open jpeg' VI to open the file, this outputs a cluster. Unbundle the image data from the cluster, which is simply a 1d array of bytes decribing the image as a raster. You can then perform numeric and array operations on the array to change the image before rebuilding the JPEG
attached is a VI I wrote to scan through a series of images in a folder, deleting similar images, and retaining images with significant difference. I know your requirement is different, but it shows how image data can be handled with simple numeric operations.