Hello All, I have taken 128MB flv video. audio is about 22 MB in size. As a very first initial step, I have converted video into image frames at 1 fps. (one frame per second). I have converted each frame into SVG image suing potrace. Resultant size of svg images is 55MB. I can further compress using diffing subsequent svg files. But there is a small problem.
Lighting and other things change from frame to frame. for example if prof has written ABC at 3rd second and ABCD at 4th second, the SVG image of ABC in 3rd frame and 4th frame is not equal. Therefore, I am not looking for "Area of movement" for example, If in 4th frame I can detect that Prof has added 'D' and If able to find rectangular area of newly inserted object then I can make a new 4th frame where unchanged area will be taken from 3rd frame and changed area will be taken from 4th frame.
What is this procedure called. I have not done any opencv or image processing task. Is there any implementation exist ?