Offline Cursive Handwritten Word Using Hidden Markov Model Technique

Abstract

Hidden Markov Model (HMM) based offline cursive manually written word segmentation technique is proposed in this strategy. In this paper, we are utilizing a classification technique to perceive the written by hand word which is SVM. Dataset collection comprises handwritten words which are in the cursive configuration images are taken as input and these pictures comprise of noise and these noises are expelled by preprocessing strategy. The preprocessing technique incorporates word picture acquisition which is an RGB image; for additional steps, the RGB image is changed over to gray image. Later, thresholding is applied to the gray image. Thinning and skeletonization is connected to the thresholded image. At that point, noise is expelled from the manually written word image and a preprocessed binary matrix appears as a matrix. Over-segmented words are partitioned by potentially segmented column (PSC) and the HMM technique. At last, the character is perceived by utilizing SVM Method.