Point density nonlinear normalized character recognition method and device

Abstract

The invention provides a point density nonlinear normalized character recognition method and a device, and the method comprises the following steps: obtaining an external rectangle of a handwritten trajectory against the handwritten trajectory of a current character inputted by a user; judging whether the width / height ratio of the external rectangle is in a preset ratio range or not; if so, expanding the external rectangle to a square, carrying out point density nonlinear normalization treatment on various pixel points in the handwritten trajectory in a two-dimensional coordinate plane of the square, and obtaining coordinates of the various pixel points after normalization; if not, directly carrying out the point density nonlinear normalization treatment on the various pixel points in the handwritten trajectory in the two-dimensional coordinate plane of the external rectangle, and obtaining the coordinates of the various pixel points after the normalization. The use of the method and the device can reduce the deformation and the distortion in character recognition and improve the reliability of the recognition.

Description

technical field

[0001] The invention relates to the technical field of handwriting recognition, in particular to a point density nonlinear normalized character recognition method and device. Background technique

[0002] In handwriting recognition, it is necessary to preprocess the Chinese characters input by the user. Since the characters are of different sizes, in order to facilitate the extraction of character features, it is necessary to normalize the size of the samples. Normalization can be divided into two methods: linear normalization and nonlinear normalization. Linear normalization is to linearly enlarge or reduce the text image according to a certain scale factor, where the scale coefficients in the X direction and the Y direction can be the same or different. The general method is to adjust each character block into a square, that is, first surround the character block with a frame, and then stretch the frame into a square. This method is mainly based on the ass...

Images