It’s summer, imagine you are on holidays in Bali. The weather is perfect, the water is clear and the sand not too hot to walk on. You want to enjoy the beach so you put on your red bikini. Now you are standing in the sand, nearby are other tourists having fun. One of them as a camera and makes snapshots of his friends. Unfortunately you are not too far away from them and appear on almost every picture, but you don’t know that. Suddenly you pay attention to the photographer and realize that you may be on some of the pictures he just took. Oh noes! You are in a very uncomfortable situation and too shy to ask the photographer to show you the pictures he just made and maybe erase the ones you don’t like. Luckily you are wearing the red P3F bikini and thanks god, the photographer uses the P3F framework to check his images for privacy policies encoded in the swimwear of the people he just recorded!
His P3F framework is just a prototype version developed in Matlab. Matlab is a great platform to develop things like that – it has cool toolboxes that make image processing easy and fun.
The core components of our Matlab implementation are:
- Preprocessor: pre-processes the image to remove distortions e.g. caused by illumination
- Face Detector: detects faces in the image
- Artefact Detector: detects P3F artefacts on people inside the image
We are currently working on a comprehensive document that describes the components in detail. We are also working on a new, enhanced version that supports more complex patterns. We are also improving face detection and body segmentation to minimize false positives.