| Summary: | 碩士 === 大同工學院 === 資訊工程研究所 === 87 === Due to the explosive growth in the multimedia application, image coding became an important and popular topic in recent years. Among the image coding techniques, the discrete cosine transform (DCT) coding is commonly used. The DCT coding is a block based coding method. The coefficients from the DCT transform are quantized to reduce the entropy, which in turn can reduce the data rate. However, the quantization error will introduce an unpleasant visual effect, the blocking effect. Blocking effect comes from the discontinuity between two adjacent blocks, which should be uniform originally.
In this thesis, we propose a new scheme to reduce the blocking effect caused by the discrete cosine transform based block coding techniques. Since it is desired not to change the original image coding scheme, we use a set of low pass filter to process the reconstructed image. However, the low pass filter may also blur the image. Therefore we must detect the original image edges to avoid it being blurred. To detect the image edges, the gradient and the thinning operator is used to get the exact edge location. Then a set of one-dimensional gaussian low pass filter is applied to the pixels at the block boundaries. We compare several previously proposed blocking effect reduction methods, and the experimental result show that this new scheme has a better performance in terms of image quality and blocking effect reduction.
|
|---|
