Investigating Jaccard Distance similarity measurement constriction on handwritten pen-based input digit
Noor Aznimah Abdul Aziz, Siti Salwa Salleh, Daud Mohamad, Megawati Omar · 2010
This paper presents a preliminary study on conventional Jaccard Distance performance in recognizing shapes with and without pre processing tasks. This study also identified the pre processing tasks that should be conducted to improve recognition performance. Jaccard Distance is performed by measuring the asymmetric information on binary variable and the comparison between vectors component. It compared two objects and notified the degree of similarity of these objects. This study also evaluated recognition performance on hand writing on isolated digits written using pen-based device. The first part of the work showed low recognition performed by the conventional Jaccard Distance when there was no pre processing task done onto the input. However, after translation, rotation, invariance scale content and noise resistance were added it showed improvement on the recognition progress. However the improvement was not very significance. Result showed that the degree of accuracy only improve averagely by 20%. Therefore thorough pre-processing tasks are to be carried out for the pen-based input using Jaccard Distance measurement.