Hi there.
I am working on a project about stitching and I considering that I know where all the images are taken, it is possible to do a rough pre-alignment. However, how to how to accelerate the stitching progress with this alignment data is a problem. The Stitcher.stitch() function only accepts the region of interests data.
https://docs.opencv.org/trunk/d2/d8d/classcv_1_1Stitcher.html#a34152e67e9b04306236b6d570e9adf30
Besides, I think it would help if I can get the alignment data generated by the stitching pipeline rather than just feeding in some images with random order and getting a stitched image as the output with almost zero control to its quality.
Does anyone have suggestions on these?