| Image interpolation is one of important contents of image processing and is an important application in digital image processing, with widely used in aerospace, medical, military, scientific research, communications, remote sensing satellite, television, film production, etc. It is an important method to improve the visual effect. And bi-cubic interpolation is an important technology in interpolation methods and has most common applications in various image interpolation technologies, because of clear output images, and without blurred edges and ladder-like phenomenon.In this paper, the bi-cubic interpolation method has been improved, and the improved bi-cubic interpolation is used in the document image interpolation, to achieve various transformations of document images such as scaling and rotation. In the bi-cubic interpolation algorithm, the floating-points are involved in the calculation, and the computational complexity is high with high calculation load, because of floating-points’ accuracy. The better the visual effect of the output image is, the higher the computational complexity is. General interpolation algorithms improve the quality and visual effect of interpolated image and get ideal interpolated image, but ignore the computational complexity. Therefore, good visual effect and low computational complexity is an important criterion for image interpolation. In this paper, the common used interpolation methods have been improved, and a specific effective computing algorithm is designed. In the interpolation process, integer arithmetic instead of floating-point, and integers are involved in the calculation. Reduce the computational complexity. Ensure the visual effect of interpolated images, and at the same time, improve the efficiency of the interpolation algorithm. Achieve the purpose of fast processing of the document image interpolation.For the computational efficiency of the interpolation algorithm and the visual effects of interpolated images, the following aspects are studies:1、The improved bi-cubic interpolation algorithm. Bi-cubic interpolation algorithm takes the16pixels around the target pixel into account. The interpolation formula is sum of these pixels gray-value multiplied by interpolation parameters. Computing values are floating-point. The computational complexity is too high. In order to improve computational efficiency, the bi-cubic interpolation algorithm has been improved in this paper, and the interpolation formula has been optimized. The improved algorithm interpolation formula is the convolution of two matrices. One is a4X4two-dimensional matrix consisting of16pixels from source image, called the pixel matrix. Another is a matrix set and each is a4X4two-dimensional matrix.2、The concept of interpolation template has been proposed. The matrix set of interpolation formula of improved bi-cubic interpolation algorithm is called interpolation template. The interpolation process is the convolution of the pixel matrix and the interpolation template. The values of the pixel matrix can be determined and are integers. In order to improve the computational speed of interpolation, the interpolation template needs to be optimized. According to the interpolated computing process, the value of interpolation template can be separately calculated and the value is calculated by coordinate mapping of pixel and new interpolation function S(w). After the interpolation template is calculated, the matching operation is done between pixel matrix and interpolation template, and the interpolation operation is done through the selected corresponding template matrix. Different image pixel matrices correspond to the specific matrix of interpolation template.3、Fast interpolation algorithm,. In the interpolation operation,98%image data is computed through the improved bi-cubic interpolation algorithm. So its computational efficiency determines the interpolation efficiency of this paper. The interpolation process of this algorithm is the convolution of the pixel matrix and the interpolation template. All of the internal element values of interpolation template and pixel matrix are integer, and the convolution process is the integer arithmetic. The computational efficiency is far greater than that of floating-point, especially in large data computing. The integer arithmetic is selected in the computational process of image interpolation, which greatly improves the computational speed of image interpolation. Therefore, the computational process of improved bi-cubic interpolation algorithm is integer arithmetic, and the computational efficiency of interpolation has been improved.4、The specific interpolation algorithm is chosen for different image regions. The regional characteristics of image contain rich information. For different research purposes, the regional characteristics of image can be specifically analyzed and the useful information is extracted. In this paper, according to the principle of image interpolation, the regional characteristics of image, which meet different interpolation conditions, are computed through the appropriate interpolation algorithm for image interpolation. |
PDF Full Text Request