I want to find the corners of a TV. That's what I'm recording a video of a TV that runs some movies, and I want to remix anything that is played on the TV on the movie. I am told that the best way to find TV corners is to add some markers (like the April tag) to the corner of the TV screen, when I record them and then find them using OpenCVI. I have not found such Lib / Method / Tag in OpenCVI (I work with Python and processing / Java). Can anyone talk to me this? I also welcome other tips on how I can recognize the screen firmly. I tried to find the edge but it is not reliable because sometimes it detects the goods based on the things inside the film. I understand that I can train my program to find an object, but I do not want to train for the calibration process because it adds a big overhead
You can write a simple algorithm for designing and identifying tags with a specific pattern. As a start, you can take a look, in which tags are identified correctly in a simplified but effective manner.
You may also be able to locate the tags without tags. You would know what is the ratio of the length and height of the TV. Use to find key lines in the image, then for each set of 4 lines, find the point of intersection of 4 lines. Measure the length of the line segment, and check that the ratio of roughly the horizontal line length to the vertical lines corresponds to the ratio of the TV's length and height. The set that gives you the closest match can be a TV. If there is not too much pattern in the background, then this algorithm will work very fast.
No comments:
Post a Comment