Study On Context-Based Reusable Techniques For Low Bit-Rate Image Transmission

Image coding and transmission over Internet or wireless channel have always been very active research areas. The judgment of a certain coding algorithm has long been the relationship between distortion and compression ratio. These efforts of image coding are guided by information theory, led to a plethora of methods. However, the state-of-the-art image coding schemes have inherent visible artifacts at low bit-rate, such as block effect introduced by DCT transform and ringing effect introduced by wavelet transform. It makes the low bit-rate-coded images difficult to be clearly displayed on a high resolution screen or magnified to check specific regions. The progresses in the vision techniques have opened a new vista in image coding. In this paper, we present an image coding scheme with primal sketch based learning, which adopts the reusable knowledge among different images.To the state-of-art compression schemes, more efforts have been put on removing spatial redundancy and temporal redundancy. Statistical redundancy removing is less popular due to the difficulty in modeling. In this paper, we propose a context-based entropy coding model which takes...