Python program to recognize english digits and alphabets using KNN algorithm.
According to these features:
- [ 0] Aspect Ratio (double) 0 < x < 10
- [ 1] Intensity of ON pixels (double) 0 < x < 1
- [ 2] Percent of ON pixels above horizontal half (double) 0 < x < 1
- [ 3] Percent of ON pixels to right of vertical half (double) 0 < x < 1
- [ 4] Percent of ON pixels on vertical line (double) 0 < x < 1
- [ 5] Percent of ON pixels on horizontal line (double) 0 < x < 1
- [ 6] Mean X of ON pixels (double) 0 < x < 1
- [ 7] Mean Y of ON pixels (double) 0 < x < 1
- [ 8] Standard Deviation X of ON pixels / Width (double) 0 < x ~< 1
- [ 9] Standard Deviation Y of ON pixels / Height (double) 0 < x ~< 1
- [10] Vertical state Flips (int) 0 =< x <= 10
- [11] Horizontal state Flips (int) 0 =< x <= 10
- [12] Is reflected Y axis (Percent) (double) 0 < x < 1
- [13] Is reflected X axis (Percent) (double) 0 < x < 1
Test Results On '120' image with Train Set of '230' image: Cells represent percent of correct matches