| 摘要 |
A system and method of processing an image including black text data and non-black text data are described. The system and method use a parametric analysis for discrimination of halftones, texts, and photographs. The parametric analysis provides results that track measurable image metrics without the inherent risk of errant decision making during classification. The creation of symbolic representations is intrinsically a classification process that is subject to error. It represents a fundamental departure from fuzzy logic image segmentation. In the parametric analysis, no pattern matching, no auto-correlation, no screening parameter calculation, and no conventional edge detector (such as a high pass filter) are used.
|
