2D Image Generation Algorithms Based On Point Cloud And Texture Data

Optical Character Recognition(OCR)technology can transfer words from document images into pure text documents automatically after recognition.It could benefit our lives tremendously.With the rapid development of computer vision technology,three-dimensional(3D)model data processing and information extraction have become an important research direction of computer image processing.To solve 3D model unable to be recognized by OCR,a new method by collecting the two-dimensional(2D)date from 3D model and reconstructing the 2D image is proposed.This algorithm effectively converts the 3D model into 2D image and realizes the function that 3D model is recognized by OCR indirectly.Firstly,some 3D model processing algorithms of the tilt correction and the distortion correction are described.Their advantages and disadvantages are given.Then the thesis comes up with the new method of the generation algorithm of 2D image based on point cloud and texture data.Next,the first step of the algorithm is the preprocessed 3D model,and then the processed model is classified based on the mesh types,which are divided into a regular grid model and an irregular model.The resolution of the target image is obtained by calculating the different grid types of 3D models.The second step is to extract pixel data of the 2D image from the texture data.The last step is to achieve the 2D image reconstruction.Finally,the reconstructive 2D image is achieved according to this paper design.The test data comparison results show that the proposed algorithm generate a corresponding 2D image by the data of 3D model effectively and realizes the function that 3D model is recognized by OCR indirectly.The OCR rate is high.