Asymmetrical Principal Component Analysis Video Coding in the Frequency Domain
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
The past few years have seen a rapid development in the area of image and video compression. With wide use of computers and consequently need for large scale storing and transmission of data therefore, efficient methods for storing and transmission of data have become an important issue of nowadays. Video compression is minimizing the number of bytes for each frame without degrading the quality of the frame. The reduction in frame size allows more frames to be transmitted through the internet and it also reduces the time required for frames to be sent over the network. Asymmetrical principal component analysis (aPCA)  and discrete wavelet transform  are two important techniques used for video compression in our implementation.In this thesis, we used aPCA for compression of facial video sequence. The idea behind aPCA is to use a part of the frame for encoding while using the entire frame for decoding. It can efficiently be used to reduce the complexity of encoding and decoding with only a slight decrease in reconstruction quality.Discrete wavelet transform decomposes the frames into subband images in different frequency domains where most of the information is stored in the low frequency subband (it is called LL). The stored information in low subband can be used for encoding in aPCA algorithm. Due to the very low amount of information which is used for encoding, the high reduction of complexity for encoding is achieved.
Place, publisher, year, edition, pages
IdentifiersURN: urn:nbn:se:umu:diva-59401OAI: oai:DiVA.org:umu-59401DiVA: diva2:552071
Master's Programme in Robotics and Control
Söderström, Ulrik, Lektor
Rönnbäck, Sven, Lektor