| Presentation | 1997/12/12 A Method for Identifying Probability Density Distribution by Gaussian Mixture for Multiresolution Image Takayuki NAKAGAKI, Yasunari YOKOTA, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Visual system is considered to multiresolve the information from outer and transmit it, and the mechanism is available for image coding. In studying the mechanism of Visual system, it is important to identify the probability density distribution of image represented by each resolution for measuring the transmission rate. In this paper, it is shown that the generalized Gaussian distribution cannot approximate exactly the probability density distribution of multiresolution image, namely, subband image. And, using Gaussian mixture distribution represented by the sum of Gaussian distribution, a method for approximating the probability density distribution of subband image is proposed. By the determination of distribution model so as to minimize the χ^2 value between the probability density distribution of subband image and the distribution model, it is shown that the Gaussian mixture distribution realize a less χ^2 value compared with the generalized Gaussian distribution even with taking the freedom of model parameter into consideration. And it is also shown that the entropy encoder designed by using the identified distribution from Gaussian mixture distribution is highly efficient for coding. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Subband coding / Generalized Gaussian distribution / Probability density distribution / χ^2 test / Gaussian mixture |
| Paper # | MBE97-116 |
| Date of Issue |
| Conference Information | |
| Committee | MBE |
|---|---|
| Conference Date | 1997/12/12(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | ME and Bio Cybernetics (MBE) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | A Method for Identifying Probability Density Distribution by Gaussian Mixture for Multiresolution Image |
| Sub Title (in English) | |
| Keyword(1) | Subband coding |
| Keyword(2) | Generalized Gaussian distribution |
| Keyword(3) | Probability density distribution |
| Keyword(4) | χ^2 test |
| Keyword(5) | Gaussian mixture |
| 1st Author's Name | Takayuki NAKAGAKI |
| 1st Author's Affiliation | Electronics and Computer Engeneering Division, Graduate school of Engeneering, Gifu University() |
| 2nd Author's Name | Yasunari YOKOTA |
| 2nd Author's Affiliation | Department of Information Science, Faculty of Engeneering, Gifu University |
| Date | 1997/12/12 |
| Paper # | MBE97-116 |
| Volume (vol) | vol.97 |
| Number (no) | 447 |
| Page | pp.pp.- |
| #Pages | 5 |
| Date of Issue | |