Handwritten digit recognition is the process of recognizing and classifying handwritten digits without human interaction. Its application field is very wide, for example Postal code recognition (Automatic sorting of mail by destination ZIP code), Digitizing hand written spreadsheets, tax forms etc. Patterns slightly shifted, distorted and even overwritten can be correctly recognized. Neural network aids in efficient recognition. A Multilayer Neural network trained with ?Backpropagation ?algorithm is used. Kirsch masks are adopted for extracting feature vectors and a multi layer clustered neural network is used for classifying numerals efficiently. The neural network will be trained with a database consisting of handwritten digits provided by writers of various ages with many different sizes and writing styles. Numerals poorly drawn or cannot be classified are rejected. A very high recognition rate, even above 90% could be obtained while using neural network.
i want basic s of pattern recognition in neural networks.some papers for research
Handwriting number recognition is a challenging problem that researchers have been bogging into for many years. The need for this arises in situations like checks in banks or recognizing numbers in car plates. There is a lot of need for a system that to let the computer understand the Arabic numbers that is written manually and also handles it like digital data. Methods like minimum distance,
decision tree and statistics have been developed by many researchers. The proposed system starts by acquiring an image containing digits, this image was digitized using some optical devices and after applying some enhancements and modifications to the digits within the image it can be recognized using feed forward back propagation algorithm.Finally it was tested and a a segmentation
method was used to fit the demands.
[b]MATERIALS AND METHODS[b]
The following are the different steps:
1)Image acquisition:The image is acquired to our
system as an input in a specific format such as bmp or jpg through the scanner or, digital camera or other digital
a sequence of preprocessing steps is appled to
be ready for the next step.
This involves reducing noise in an image. In the offline mode, the noise may come from the
writing style or from the optical device captures but in the online mode, there is no noise to eliminate so no need for the noise
This is done to standardize the font size
within the image.
5)Thinning and skeletonization:
This is done for Representing the shape
of the object in a relatively smaller number of pixels.
Since the data are isolated, no need for
7)Normalization scaling and translation:
The variability in size of written digits leads
to the need of scaling the digits size within the image to
a standard size.
This is done to extract the Structural features, Statistical features, and Global transformation features from the image.
Classification and recognition:
popular architecture Of Neural Networks used in
Arabic digits recognition takes a network with three
layers viz the Input layer, hidden layer and output
layer.The nodes in the input layer varies according to the feature vectorâ„¢s dimensionality of the
segment image size.
Generally the neural networks are trained so that a particular input leads to a specific target output. Then the network is adjusted, based on a comparison of the output and the target. This is a feed forward back propagation.
for full details, refer this pdf: