An elastic matching approach applied to digit recognition

Abstract

Usually used methodr for digit recognition on handwritten mail addresses are either structural or statistical. The structural ones operate by expert rule application. When the extraction process of such rules is done by hand, these approaches are generally dijjicult to design. The statistical methodr use &generated image descriptions of the digit which consist mainly of feature vectors. The decision zone learning in the corresponding vectorial space allows vector clwerization. In such techniques, the system recognition result is directly related to the chosen measure vector relevance. This choice is dfjcult and no completely satisfactory solution exists in this domain. In order to overcome the drawbackc of these methods, we have evaluated an elastic matching algorithm. This approach models and quantifies the distorswns undergone by the digit to be recognized as compared with the theoretical and ideal pattern.Consequently, there is neither obligation of a decision rule elaboration nor a need for feature vector design. We describe the principles of our elastic matching algorithm and analyses its performance when ran on a reference set.

On deformable models for visual pattern recognition

“…The domains include human faces [27,58,101,127,128], gestures [50,72], handwriting [1,4,26,54,81,104,129,130], etc. See Fig.…”

Using generative models for handwritten digit recognition

“…This advantage of spline models is pointed out in [35] where a different kind of spline is used to fit on-line character data by directly locating candidate control points on strokes in the image. It is lost (as pointed out in [36]) when models based more directly on Durbin and Willshaw's elastic net are employed [37].…”