Font Size: a A A

A Study Of Inverse Halftoning And Quality Assessment Schemes

Posted on:2009-03-03Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y P KongFull Text:PDF
GTID:1118360245968510Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Digital Halftoning and inverse halftoning is one of the supported techniques to the image input and output for computer, and it is also a marginal branch in image processing field. By using halftoning technique, digital images with n-level grayscale are converted into special binary images, and the conflict that some computer peripherals only can display and print two tones is solved. However, halftoning process not only introduces the halftoning noise but also changes the image characteristics of tones. So the applications of image zoom, enhancement, compression and recognition can not be done on halftoning images directly, and the processing of inverse halftoning is needed to recover a con-tone image from its halftone version. Because halftoning and inverse halftoning both appear one to more mapping, which makes inverse halftoning become an ill-posed problem, the inverse halftoning and its quality assessment techniques have become important subjects in the image input and output field.In this dissertation, the basic principle, systemic models, typical methods and classifications of digital halftoning are summarized firstly. Then the characteristics of halftones in spatial and frequency domains are analyzed. Analogously, the technical outlines, mathematic models, typical methods and its advantages and disadvantages of existing inverse halftoning methods are summarized. After that, some pivotal techniques are researched such as halftoning pattern recognition, inverse halftoning for grey and color image and its quality assessment.By the analysis of all kinds of existing inverse halftoning schemes, it is found that they either need enough prior knowledge of halftone or only used for some typical halftones, which makes the validity, adaptability and flexibility of the schemes at a discount state. So the recognition technique for halftone image classification is studied first. And a recognition algorithm based on one-dimensional correlation and texture characteristic is proposed, which meets the needs for supporting and optimizing inverse halftone schemes and defining its application range.Meanwhile, aiming at the problems on the limitation of reconstructed image quality, three new inverse halftoning algorithms for gray image are proposed, among which some methods make the new algorithms fast and effective such as Elman recurrent network, B-spline wavelet transformation, median interpolating pyramid transformation, self-adaptive filter and texture direction controlling.On the other hand, aiming at the color inverse halftoning the Human Visual System (HVS), computation model for color, orthogonal transformation, pattern-color separable model, color noise detection and denoising are studied in the dissertation. It offers three new inverse halftone algorithms for color images which can redisplay con-tone image with less color deviation and higher accuracy.Because the usual image quality assessment methods such as Peak Signal-to-Noise Ratio (PSNR), Mean Square Error(MSE) and CIELAB color difference can not correspondingly describe the suppression level to halftoning noise, artificial texture and color difference of inverse halftoning result, and three new quality assessment algorithms for inverse halftoning image are proposed through multi-perspective research such as color sensory characteristics of HVS, brightness and chromaticity separable model, simulation filtering of HVS, extended spatial of CIELAB, corrected formula of color difference, reference threshold for sensory color difference, visualization representation, etc. Among them, the results of new algorithms can describe the state of edge protecting, halftoning texture suppression and color difference of perception. It has been changed that there is no method to assess the quality of inverse halftoning fully and specifically.Finally, the following pended issue in this research field is discussed, and lots of directions are given.
Keywords/Search Tags:Image Processing, Digital Halftoning, Inverse Halftoning, Image Quality Assessment
PDF Full Text Request
Related items