Learning based screen image compression
There are usually two components in computer screen images: textual and pictorial parts. The pictorial part can be compressed efficiently by classical coding approaches (e.g. JPEG, JPEG2000), while the compression of the textual part is still far away from being satisfactory for the reason that the...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/96497 http://hdl.handle.net/10220/11929 |
| _version_ | 1826120727269474304 |
|---|---|
| author | Yang, Huan Lin, Weisi Deng, Chenwei |
| author2 | School of Computer Engineering |
| author_facet | School of Computer Engineering Yang, Huan Lin, Weisi Deng, Chenwei |
| author_sort | Yang, Huan |
| collection | NTU |
| description | There are usually two components in computer screen images: textual and pictorial parts. The pictorial part can be compressed efficiently by classical coding approaches (e.g. JPEG, JPEG2000), while the compression of the textual part is still far away from being satisfactory for the reason that the textual content is usually of high-frequency. In this paper, a learning approach is used to construct a tailored dictionary for text representation. Based on the learned dictionary, a novel screen image compression algorithm is proposed through adopting different basis functions for the textual and pictorial components respectively. The screen images are firstly segmented into textual and pictorial parts. Then we employ traditional discrete cosine transformation (DCT) to facilitate the compression of pictorial part, while the learned dictionary is used to represent the textual part in screen images. Experimental results demonstrate the effectiveness of the proposed compression algorithm. |
| first_indexed | 2024-10-01T05:21:09Z |
| format | Conference Paper |
| id | ntu-10356/96497 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:21:09Z |
| publishDate | 2013 |
| record_format | dspace |
| spelling | ntu-10356/964972020-05-28T07:18:05Z Learning based screen image compression Yang, Huan Lin, Weisi Deng, Chenwei School of Computer Engineering IEEE International Workshop on Multimedia Signal Processing (14th : 2012 : Banff, Alberta, Canada) DRNTU::Engineering::Computer science and engineering There are usually two components in computer screen images: textual and pictorial parts. The pictorial part can be compressed efficiently by classical coding approaches (e.g. JPEG, JPEG2000), while the compression of the textual part is still far away from being satisfactory for the reason that the textual content is usually of high-frequency. In this paper, a learning approach is used to construct a tailored dictionary for text representation. Based on the learned dictionary, a novel screen image compression algorithm is proposed through adopting different basis functions for the textual and pictorial components respectively. The screen images are firstly segmented into textual and pictorial parts. Then we employ traditional discrete cosine transformation (DCT) to facilitate the compression of pictorial part, while the learned dictionary is used to represent the textual part in screen images. Experimental results demonstrate the effectiveness of the proposed compression algorithm. 2013-07-22T03:09:06Z 2019-12-06T19:31:27Z 2013-07-22T03:09:06Z 2019-12-06T19:31:27Z 2012 2012 Conference Paper Yang, H., Lin, W., & Deng, C. (2012). Learning based screen image compression. 2012 IEEE 14th International Workshop on Multimedia Signal Processing (MMSP). https://hdl.handle.net/10356/96497 http://hdl.handle.net/10220/11929 10.1109/MMSP.2012.6343419 en © 2012 IEEE. |
| spellingShingle | DRNTU::Engineering::Computer science and engineering Yang, Huan Lin, Weisi Deng, Chenwei Learning based screen image compression |
| title | Learning based screen image compression |
| title_full | Learning based screen image compression |
| title_fullStr | Learning based screen image compression |
| title_full_unstemmed | Learning based screen image compression |
| title_short | Learning based screen image compression |
| title_sort | learning based screen image compression |
| topic | DRNTU::Engineering::Computer science and engineering |
| url | https://hdl.handle.net/10356/96497 http://hdl.handle.net/10220/11929 |
| work_keys_str_mv | AT yanghuan learningbasedscreenimagecompression AT linweisi learningbasedscreenimagecompression AT dengchenwei learningbasedscreenimagecompression |