| 摘要 |
Traditional techniques such as filtering and edge enhancement have been applied to restoring images that have been distorted due to lossy image compression. However, these techniques have ignored a unique feature that can be exploited when working with digital compression. Before the image is stored or transmitted, the sender has access to both the original and the distorted images, enabling the encoder to transmit information specifying the regions where the enhancement was successful. To utilize this feature, before storage or transmission the sender produces a codec file, and assesses the efficacy of one or more enhancement schemes. To determine which image regions have been improved by the enhancement, the enhanced codec is compared to the original. A map of where the enhancement scheme was successful is encoded into the image by making tiny adjustments to the image itself. This method can comply with many compression standards, adds nothing to the compressed images' bandwidth, and distorts the image by only a very tiny amount. Examples are presented showing the reduction of distortion if enhancement is used, together with calculations of how much distortion is introduced if a standard decompressor is used.
|
