EXISTINGSYSTEM:Inthe exiting system of the gender prediction from handwriting there are somelimitations. A computer is not allowed to transcript the content of difficulthandwritten document. While writing there are many difficulties are produced byinterpersonal and intrapersonal. Noisy background is also a limitation ofexisting system. The supervised learning problems are considered as binary ormulti-class ones. Multi-class problems, discrimination in different age, humanraces, and different nationalities. PROPOSEDSYSTEM:Theproposed system presents the study to predict gender of individuals from the scannedcopy of their handwritings.
The proposed system is based on extracting the setof features from writing samples of male and female writers and trainingclassifiers to learn to differentiates between the two. Images are differentiatedusing Otsu thresholding algorithm. The following features have been considered.Writing attributes like slant, curvature, texture and liability are estimatedby computing local and global features. Classification is done using artificialneural network and support vector machine. Support vector machine are the setsof related supervised learning which can be used for both classification andregression.
In simple word, an SVM classification tries to build a decisionmodel capable of predicting one category falls into the other. Support vectormachine classifies the images on test dataset. SVM is used to compute hyperplanewith maximum margin. The computation for the output of a given SVM, Artificial neural network, the sequencerecognition is used for handwriting detection.
The proposed technique evaluatedon the databases resulting the gender prediction from handwriting.Some features which is used are:· Tortuosity: this is usedto differential between the smooth handwriting and twisted handwriting.· Direction: this is usedto measure tangent direction of text.· Curvatures.· Chain code.· Edge direction.