Abstract
In this paper, we present a new feature extraction technique called ”ERP”. For Transformation-ring-projection (TRP) method, it loses indispensable information by reducing the dimensionality of a two dimensional pattern. In particular, it is not rotation invariant in all cases. For example, it has identical classified curves for mirror characters like ”人” (man) and ”入” (enter). In an effort to correct this deficiency, ERP is developed to enhance the TRP method.
The advantages of proposed ERP method in this problem domain are:
i) It keeps enough information to distinguish mirror characters during transformation from a two dimensional pattern.
ii) It is authentic rotation invariant.
In pattern recognition, the entire task can be divided into four phases, namely, data acquisition, data processing, feature extraction and classification.
The advantages of proposed ERP method in this problem domain are:
i) It keeps enough information to distinguish mirror characters during transformation from a two dimensional pattern.
ii) It is authentic rotation invariant.
In pattern recognition, the entire task can be divided into four phases, namely, data acquisition, data processing, feature extraction and classification.
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the 4th IAPR International Workshop on Document Analysis Systems, DAS 2000 |
| Pages | 197-210 |
| Number of pages | 14 |
| Publication status | Published - 10 Dec 2000 |
| Externally published | Yes |
| Event | 4th IAPR International Workshop on Document Analysis Systems, DAS 2000 - Rio Othon Palace Hotel, Rio de Janeiro, Brazil Duration: 10 Dec 2000 → 13 Dec 2000 |
Conference
| Conference | 4th IAPR International Workshop on Document Analysis Systems, DAS 2000 |
|---|---|
| Country/Territory | Brazil |
| City | Rio de Janeiro |
| Period | 10/12/00 → 13/12/00 |
User-Defined Keywords
- Enhanced-ring-projection (ERP)
- character group
- the local maximal modulus of wavelet transform and unique pattern scheme