A Dimensionality Reduction Method for Handwritten Characters in 3D Space Based on Longest Trajectory Projection

Abstract

The invention discloses a longest track projection-based three-dimensional space handwriting character dimension reduction method. A conventional dimension reduction method has defects in the dimension reduction aspect of three-dimensional handwriting character identification. The method sequentially comprises the following steps of obtaining three-dimensional coordinates of a moving fingertip; generating a three-dimensional track; projecting the three-dimensional coordinates to three standard planes; connecting two-dimensional coordinate points in sequence to generate two-dimensional tracks;calculating the length of the two-dimensional track on each plane; deleting the length of an overlapping part in the two-dimensional tracks; then comparing the lengths of the three two-dimensional tracks obtained in different planes and selecting out a longest track; and defining the plane where the longest track is located as an optimal projection plane, wherein the two-dimensional projection track in the optimal projection plane is a two-dimensional handwriting character subjected to dimension reduction. A 2D image in a fixed direction can be obtained; the calculation and storage costs can be reduced; a relatively good visual effect can be obtained; and in the three-dimensional handwriting character identification, direction adjustment does not need to be carried out, so that the three-dimensional space handwriting character identification rate is increased.

Description

Technical field [0001] The invention belongs to the technical field of dimensionality reduction, and specifically relates to a method for dimensionality reduction of three-dimensional space handwritten characters based on the longest trajectory projection. Background technique [0002] In many practical applications, such as machine learning, image classification, image processing and pattern recognition, the dimensionality of the original data is always high. High-dimensional data requires high memory, and it is troublesome to process, and computationally expensive. This problem is the so-called "curse of dimensionality." To solve this problem, many dimensionality reduction methods have been proposed. So far, the traditional dimensionality reduction methods still have some shortcomings and shortcomings. The disadvantage of Principal Component Analysis (PCA) is that the direction of the 2D characters after dimensionality reduction is random, and better visual effects cannot be...

AI Technical Summary

Problems solved by technology

So far, traditional dimensionality reduction methods still have some shortcomings and deficiencies

The disadvantage of principal component analysis (PCA) is that the direction of 2D characters after dimensionality reduction is random and cannot obtain better visual effects; in the recognition of handwritten characters in three-dimensional space, it is necessary to adjust the direction of the 2D image before image recognition. Otherwise it will increase the misrecognition rate of 2D images

The limitation of linear discriminant analysis (LDA) is that the data after dimension reduction can only be projected to C-1 dimension at most (C is the number of categories of the original data), and the handwritten characters belong to one category, so LDA cannot be used for three-dimensional handwriting Dimensionality reduction of characters