Handwritten Character Recognition Using A Gradient Classifier
R.D. Brandt, Yao Wang, Alan J. Laub, Sanjit K. Mitra · 2005
We consider a prototype-based character recognizer that makes comparisons based on blurred representations of the images. The blurring induces a metric on the space of all images that varies continuoitsly under continuous deformations of the image plane. This blurred representation is suitable for direct implementation of a nearest neighbor classifier. However, it is still desirable to have a representation which is invariant under rotation, translation and scaling of the image plane. A representation which is locally invariant under these transformations is produced by transforming an input to a local minimum of its distance from each prototype simultaneously. These minima are found by performing a gradient descent on an appropriate error surface over the 4 transformation parameters. The error functional is the L/sub 2/-norm of the difference between the blurred prototype and the blurred input. The resulting classifier makes more efficient use of prototypes than the nearest neighbor classifier.