Releases: ProjectAida/aida
v0.3.0
Third pre-release of aida code, including improved page segmentation and improved pre-processing strategies.
Note: the current strategy of new image processing approach are:
--- A. Binarization
----- 1. Otsu's method is adopted.
--- B. Consolidation
----- 1. The overall strategy is "mixed" which combines non-recursion and recursion results using OR operator.
----- 2. Horizontal Projection for Row Histogram obtains: using the average number of textual pixels as the threshold.
----- 3. Common Block Heights Extraction: using the median line height as the threshold.
----- 4. Text Line Consolidation for Textual Line smearing: using the average percentage of all regions as the threshold.
v0.2.0
This version, v0.2.0, includes new page segmentation.
In this version, page segmentation is able to segment 2-column layout newspaper image.
v0.1.0
First pre-release of aida code, as developed under/for NEH Digital Humanities Start-up Grant, HD-51897-14