Digital image compression by using edge based segmentation /
The purpose of digital image compression is to reduce the volume of visual information in the encoding/transmitting end such that the reconstructed image in the decoding/receiving end will possess small or even no distortion. The art of designing such an image compression system is in balancing the...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1996.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=739667471&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | The purpose of digital image compression is to reduce the volume of visual information in the encoding/transmitting end such that the reconstructed image in the decoding/receiving end will possess small or even no distortion. The art of designing such an image compression system is in balancing the compression ratio, the distortion and the processing time. In this dissertation, we have implemented a new coding scheme to compress a gray-scale image for human visual interest. To gain a higher compression ratio with a smaller distortion, the properties of the human visual system have been exploited. An important visual feature (i.e., the edge point) is utilized in the segmentation of a given image. This image segmentation procedure confines the detected edge points into small square regions and partitions the remaining areas into larger square sub-images. According to the segmentation results, Adaptive Vector Quantization (AVQ) technique is used to compress the square sub-images that contain edge points, and Slant Plane Approximation (SPA) method is employed to code the larger square sub-image blocks that contain no edge points. Thus, the characteristics of the different types of sub-images can be better exploited. Owing to the consideration of processing efficiency and the innate technical limitation, the AVQ and SPA techniques cannot adequately compress a given image. The remaining image information, which is the difference between the given image and the AVQ as well as the SPA techniques coded image, is then compressed by Discrete Cosine Transform (DCT) coding technique. Therefore, a finer decompressed image can be obtained in the decoding/receiving end. The final result shows that the reconstructed image which is processed by our proposed compression scheme will have a better image resolution than that coded only by the DCT technique under the same bit rate condition. In the investigation of noise effects, our proposed scheme can endure moderate noise and preserve perceptible image information in the decompressed image. |
|---|---|
| Item Description: | Vita. "Major Subject: Electrical Engineering". |
| Physical Description: | xiv, 93 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilms Inc. |
| Bibliography: | Includes bibliographical references: pages 88-91. |