Onlinehandwriting analysis is a challenging research area for the last few decades. Torecognize with perfect prediction some pre-processing steps are essential. Inthis paper one of the important pre-processing steps dehooking is presented. Herewe consider Bengali online handwritten Characters and words as sample for removinghooks.
Hooks are basically common artifacts during fast writing by people.Hooks are very common issue present at the beginning in very rare case and theend of character stroke in maximum case and are generated by the pen-down andpen up movements respectively. Dehooking is the process of eliminating suchunwanted strokes that appear due to inaccuracies in pen down position.Dehooking algorithms are applied to remove hooks. Here, strokes are detected bycomparing the number of points with a threshold value.
If the value is greaterthan the threshold value, the mark is retained or it is removed otherwise. Inthis new and innovative approach we focus on the dehooking at the end ofcharacter stroke and consider last 20 percent of each stroke for checking,according to distance from the co-ordinate of the first pixel. In last 20percent of a stroke, we calculated angle among three consecutive pixels. If ina particular point, angle among three consecutive pixels is falling suddenlythen immediately we pointed out that point. After pointing out the anglefalling place we checked the entire remaining pixel after that point, whetherall the remaining points are getting fade slowly or not. If it is found thatall the remaining points aregetting fade slowly then it can be assumed that ita hook.
After detecting the hook for a particular stroke we removed all theremaining pixels from the falling angle place so that hook can be removed andthe handwritten character will be hook less.I tested 1200 English and 1600Bengali online handwritten characters and we got 97.02 percent of accuracy.