SRT project of evaluating and optimizing Chinese characters calligraphy. This project includes loading the character as monochronic graph to 0-1 matrix and refine it to get better pattern. Futhermore, we feed these patterns to CNN to get the best range (interval) of these parameters.
- general.py
- imgproc.py
- carwler/catch.py
- list_proc.py
| Name | Meaning |
|---|---|
| two_threshold | Threshold for transforming a RGB picture into a black-white picture |
| Function | Argument | Illustration |
|---|---|---|
| imgread | file address | read pictrue(jpg), and transform it into a matrix(2d) |
| imgshow | picture matrtix | show picture, press any key to quit |
| matread | pictrue matrix | return a num matrix construct by 0 and 1, 0 mean white,1 mean black |
| matwrite | number matrix | return a pictrue matrix |
| matget | file address | read pictrue, and return a num matrix directly |
| pos_black | number matrix | get all black pot, and return a list with all postion[x,y] of black point |
| convexhull | picture or number matrix | return the turning point of the convex hull |
| drawconvex | picture matrix,contours | draw convex into pictrue matrix |
| findcenter | matrix, isimg(bool)=False | when isimg is False,this function accept num matrix(0,1),yet it accept image matrix(0,255) |
| findaxis | number matrix | return the axis of the black point calculated by linear regression |
| matclear | number matrix | clear up the little point in a matrix |