Research Of Font Standard Compliance Detection Technology

Due to the development of computer technology, information technology, network technology, embedded technology and others related technologies, fonts are more widely used on the computer and mobile phone-related electronic products. The font production process should be completed in accordance with certain criteria, and to comply with the requirements of the countries and the relevant international organizations in the glyph and encoding. According to character encoding of the standard font to extract the character images whose coding is corresponding to it and the same kinds of fonts (for example, are white body) from standard font and measured font, then corresponding comparison of these images and judging whether it meets the glyph of standard font or not, are the focus of my thesis. Detecting the glyph of font was made by someone is whether or not in line with the glyph of national and relevant international organizations’standard font, that is, whether the character glyph of the measured font is matched with the character glyph of the standard font.According to the IT standards of Mongolian, Chinese, English, and the font file specification describes of True Type, the research content of font standard compliance detecting technology is determined, and the corresponding languages font standard conformance test contents and scheme are proposed as well. Detecting method and its realization for the glyph of different languages font is our main research goal. How to choose a more optimal glyph feature detection and matching method, and get a better matching results on the reasonable domain value, is our concern.First, extract features of the glyph from the character image, and then calculate similarity about the glyph features. In the case of a different threshold, adjusting the parameters, get the more optimized values of similarity. By means of analyzing the nature of the character image extracted from the font, a combination of feature extraction method of a shape-based PHOG (gradient direction histogram pyramid) characteristics and the Shape Context feature is proposed in this thesis, which not only combines the advantages of PHOG and Shape Context, but also complement each other’s shortcomings. Compared with wavelet pyramid energy distribution character and SIFT, the advantage and better experimental of the integrated approach of this paper is reached. Besides, similarity distance is calculated by x2the distance method, histogram method and the minimum cost matrix method with the above extraction characteristics corresponding combining. Experiments show that our method has a better effect than those only using any one of them, improving the match degree and speed of operation about character image processing.