Codage vidéo distribué pour les images médicales application aux capsules endoscopiques.

Abstract

Distributed video coding (DVC) is a new compression paradigm that enables less complex, robust and less latency encoding for some applications. Traditional video coding contains an encoder 5 to 10 times more complex than the decoder and requires more energy and hardware resources for coding. However, distributed coding is a candidate for wireless applications that contain limited energy sources. The wireless capsule endoscopy is a medical tool introduced in 2000 and used by doctors in the diagnosis of gastrointestinal diseases. This tool cannot visualize the completely gastrointestinal system because the battery life is limited. One of the proposed solutions is to use a simple and efficient compression system to reduce the number of bits transmitted by the antenna that consume the most energy from the battery. In this thesis, we propose new distributed video coding systems simple, efficient and applicable for the images of the capsule endoscopy. The proposed DVC systems are a low complexity because uses a single type of frame compared to traditional DVC systems that use two types of frames. Both proposed DVC systems use the database to generate the side information (SI) that used to decode the source frame. The selection made in the database using the coefficients DC in the first contribution and the subsample image in the second. For the third contribution, the codec uses the vector quantization to generate the side information (SI). Several algorithms are implemented in the proposed systems to improve compression efficiency. The results obtained are significant and prove the ideas proposed by comparing them with the Motion JPEG codec already used in the endoscopic capsule.