<p>This report describes the procedure in which we accomplished our task, which was to implement a present algorithm. This was made using C++. The algorithm is developed by Sasan Gooran, who is our examiner in this project. The algorithm uses an iterative method to produce the halftoned images. </p><p>The reason for this project is to speed up the existing program which is made in Matlab. Our main action taken to speed up the program was to use C++ instead of Matlab. It is safe to say that we managed to fulfill our commitments with respect to speed, but we are not completely satisfied with the results regarding the quality. Although the result does not differ a great deal compared to the Matlab version, there are still improvements to make before considering using it commercially. </p><p>Our program is an extension of an existing C++ program, written by Per Fredriksson. We used his program as base to our program. The addition we have made to his code is the possibility to halftone colour images.</p>

With the advancement of display technology, consumers expect high quality display of image and video data. Many viewers are used to watching video content on high definition television and large screens. However, certain display technologies, such as several of those used in portable electronic books, are limited on resources such as the availability of number of bits per pixel (i.e. the bit-depth). Display of good or even acceptable perceptual quality video on these devices is a hard technical problem that a display designer must solve. Video halftoning reduces the number of represented colors or gray levels for display on devices that are unable to render the video at full bit-depth. Bit-depth reduction results in visible spatial and temporal artifacts. The designer would want to choose the halftoning algorithm that reduces these artifacts while meeting the target platform constraints. These constraints include available bit-depth, spatial resolution, computational power, and desired frame rate. Perceptual quality assessment techniques are useful in comparing different video halftoning algorithms that satisfy the constraints. This dissertation develops a framework for the evaluation of two key temporal artifacts, flicker and dirty-window-effect, in medium frame rate binary video halftones generated from grayscale continuous-tone videos. The possible causes underlying these temporal artifacts are discussed. The framework is based on perceptual criteria and incorporates properties of the human visual system. The framework allows for independent assessment of each of the temporal artifacts. This dissertation presents design of algorithms that generate medium frame rate binary halftone videos. The design of the presented video halftone generation algorithms benefits from the proposed temporal artifact evaluation framework and is geared towards reducing the visibility of temporal artifacts in the generated medium frame rate binary halftone videos. This dissertation compares the relative power consumption associated with several medium frame rate binary halftone videos generated using different video halftone generation algorithms. The presented power performance analysis is generally applicable to bistable display devices. This dissertation develops algorithms to enhance medium frame rate binary halftone videos by reducing flicker. The designed enhancement algorithms reduce flicker while attempting to constrain any resulting increase in perceptual degradation of the spatial quality of the halftone frames. This dissertation develops algorithms to enhance medium frame rate binary halftone videos by reducing dirty-window-effect. The enhancement algorithms reduce dirty-window-effect while attempting to constrain any resulting increase in perceptual degradation of the spatial quality of the halftone frames. Finally, this dissertation proposes design of medium frame rate binary halftone video enhancement algorithms that attempt to reduce a temporal artifact, flicker or dirty-window-effect, under both spatial and temporal quality constraints. Temporal quality control is incorporated by using the temporal artifact assessment framework developed in this dissertation. The incorporation of temporal quality control, in the process of reducing flicker or dirty-window-effect, helps establish a balance between the two temporal artifacts in the enhanced video. At the same time, the spatial quality control attempts to constrain any increase in perceptual degradation of the spatial quality of the enhanced halftone frames. / text

Digital halftoning is a crucial technique used in digital printers to convert a continuoustone image into a pattern of black and white dots. Halftoning is used since printers have a limited availability of inks and cannot reproduce all the color intensities in a continuous image. Error Diffusion is an algorithm in halftoning that iteratively quantizes pixels in a neighborhood dependent fashion. This thesis focuses on the development and design of a parallel scalable hardware architecture for high performance implementation of a high quality Stacked Error Diffusion algorithm. The algorithm is described in ‘C’ and requires a significant processing time when implemented on a conventional CPU. Thus, a new hardware processor architecture is developed to implement the algorithm and is implemented to and tested on a Xilinx Virtex 5 FPGA chip. There is an extraordinary decrease in the run time of the algorithm when run on the newly proposed parallel architecture implemented to FPGA technology compared to execution on a single CPU. The new parallel architecture is described using the Verilog Hardware Description Language. Post-synthesis and post-implementation, performance based Hardware Description Language (HDL), simulation validation of the new parallel architecture is achieved via use of the ModelSim CAD simulation tool.

This report describes the procedure in which we accomplished our task, which was to implement a present algorithm. This was made using C++. The algorithm is developed by Sasan Gooran, who is our examiner in this project. The algorithm uses an iterative method to produce the halftoned images. The reason for this project is to speed up the existing program which is made in Matlab. Our main action taken to speed up the program was to use C++ instead of Matlab. It is safe to say that we managed to fulfill our commitments with respect to speed, but we are not completely satisfied with the results regarding the quality. Although the result does not differ a great deal compared to the Matlab version, there are still improvements to make before considering using it commercially. Our program is an extension of an existing C++ program, written by Per Fredriksson. We used his program as base to our program. The addition we have made to his code is the possibility to halftone colour images.

<p>In this thesis, studies on three problems in imaging systems will be discussed.</p> <p>The first problem deals with re-rendering segments of online indoor room images with preferred textures through websites to try new decoration ideas. Previous methods need too much manual positioning and alignment. In the thesis, a novel approach is presented to automatically achieve a natural outcome with respect to indoor room geometry layout.</p> <p>For the second problem, the laser electrophotographic system is eagerly looking for a digital halftoning algorithm that can deal with unequal printing resolution, since most halftoning algorithms are focused on equal resolution. In the thesis, a novel monochrome halftoning algorithm is presented to render continuous tone images with limited numbers of tone levels for laser printers with unequal printing resolution.</p> <p>For the third problem, a novel face set recognition method is presented. Face set recognition is important for face video analysis and face clustering in multiple imaging systems. And it is very challenging considering the variation of image sharpness, face directions and illuminations for different frames, as well as the number and the order of images in the face set. To tackle the problem, a novel convolutional neural network system is presented to generate a fixed-dimensional compact feature representation for the face set. The system collects information from all the images in the set while having emphasis on more frontal and sharper face images, and it is regardless of the number and the order of images. The generated feature representations allow direct, immediate similarity computation for face sets, thus can be directly used for recognition. The experiment result shows that our method outperforms other state of-the-art methods on the public test dataset.</p>

Halftones are reproductions of images created through the process of halftoning. The goal of halftones is to create a replica of an image which, at a distance, looks nearly identical to the original. Several different methods for producing these halftones are available, three of which are error diffusion, DBS and IMCDP. To check whether a halftone would be perceived as of high quality there are two options: Subjective image quality assessments (IQA’s) and objective image quality (IQ) measurements. As subjective IQA’s often take too much time and resources, objective IQ measurements are preferred. But as there is no standard for which metric should be used when working with halftones, this brings the question of which one to use. For this project both online and on-location subjective testing was performed where observers were tasked with ranking halftoned images based on perceived image quality, the images themselves being chosen specifically to show a wide range of characteristics such as brightness and level of detail. The results of these tests were compiled and then compared to that of eight different objective metrics, the list of which is the following: MSE, PSNR, S-CIELAB, SSIM, BlurMetric, BRISQUE, NIQE and PIQE. The subjective and objective results were compared using Z-scores and showed that SSIM and NIQE were the objective metrics which most closely resembled the subjective results. The online and on-location subjective tests differed greatly for dark colour halftones and colour halftones containing smooth transitions, with a smaller variation for the other categories chosen. What did not change was the clear preference for DBS by both the observers and the objective IQ metrics, making it the better of the three methods tested. / <p>Examensarbetet är utfört vid Institutionen för teknik och naturvetenskap (ITN) vid Tekniska fakulteten, Linköpings universitet</p>

Le présent mémoire comprend un survol des principales méthodes de rendu en demi-tons, de l’analog screening à la recherche binaire directe en passant par l’ordered dither, avec une attention particulière pour la diffusion d’erreur. Ces méthodes seront comparées dans la perspective moderne de la sensibilité à la structure. Une nouvelle méthode de rendu en demi-tons par diffusion d’erreur est présentée et soumise à diverses évaluations. La méthode proposée se veut originale, simple, autant à même de préserver le caractère structurel des images que la méthode à l’état de l’art, et plus rapide que cette dernière par deux à trois ordres de magnitude. D’abord, l’image est décomposée en fréquences locales caractéristiques. Puis, le comportement de base de la méthode proposée est donné. Ensuite, un ensemble minutieusement choisi de paramètres permet de modifier ce comportement de façon à épouser les différents caractères fréquentiels locaux. Finalement, une calibration détermine les bons paramètres à associer à chaque fréquence possible. Une fois l’algorithme assemblé, toute image peut être traitée très rapidement : chaque pixel est attaché à une fréquence propre, cette fréquence sert d’indice pour la table de calibration, les paramètres de diffusion appropriés sont récupérés, et la couleur de sortie déterminée pour le pixel contribue en espérance à souligner la structure dont il fait partie. / This work covers some important methods in the domain of halftoning: analog screening, ordered dither, direct binary search, and most particularly error diffusion. The methods will be compared in the modern perspective of sensitivity to structure. A novel halftoning method is also presented and subjected to various evaluations. It produces images of visual quality comparable to that of the state-of-the-art Structure-aware Halftoning method; at the same time, it is two to three orders of magnitude faster. First is described how an image can be decomposed into its local frequency content. Then, the basic behavior of the proposed method is given. Next, a carefully chosen set of parameters is presented that allow modifications to this behavior, so as to maximize the eventual reactivity to frequency content. Finally, a calibration step determines what values the parameters should take for any local frequency information encountered. Once the algorithm is assembled, any image can be treated very efficiently: each pixel is attached to its dominant frequency, the frequency serves as lookup index to the calibration table, proper diffusion parameters are retrieved, and the determined output color contributes in expectation to underline the structure from which the pixel comes.

Le présent mémoire comprend un survol des principales méthodes de rendu en demi-tons, de l’analog screening à la recherche binaire directe en passant par l’ordered dither, avec une attention particulière pour la diffusion d’erreur. Ces méthodes seront comparées dans la perspective moderne de la sensibilité à la structure. Une nouvelle méthode de rendu en demi-tons par diffusion d’erreur est présentée et soumise à diverses évaluations. La méthode proposée se veut originale, simple, autant à même de préserver le caractère structurel des images que la méthode à l’état de l’art, et plus rapide que cette dernière par deux à trois ordres de magnitude. D’abord, l’image est décomposée en fréquences locales caractéristiques. Puis, le comportement de base de la méthode proposée est donné. Ensuite, un ensemble minutieusement choisi de paramètres permet de modifier ce comportement de façon à épouser les différents caractères fréquentiels locaux. Finalement, une calibration détermine les bons paramètres à associer à chaque fréquence possible. Une fois l’algorithme assemblé, toute image peut être traitée très rapidement : chaque pixel est attaché à une fréquence propre, cette fréquence sert d’indice pour la table de calibration, les paramètres de diffusion appropriés sont récupérés, et la couleur de sortie déterminée pour le pixel contribue en espérance à souligner la structure dont il fait partie. / This work covers some important methods in the domain of halftoning: analog screening, ordered dither, direct binary search, and most particularly error diffusion. The methods will be compared in the modern perspective of sensitivity to structure. A novel halftoning method is also presented and subjected to various evaluations. It produces images of visual quality comparable to that of the state-of-the-art Structure-aware Halftoning method; at the same time, it is two to three orders of magnitude faster. First is described how an image can be decomposed into its local frequency content. Then, the basic behavior of the proposed method is given. Next, a carefully chosen set of parameters is presented that allow modifications to this behavior, so as to maximize the eventual reactivity to frequency content. Finally, a calibration step determines what values the parameters should take for any local frequency information encountered. Once the algorithm is assembled, any image can be treated very efficiently: each pixel is attached to its dominant frequency, the frequency serves as lookup index to the calibration table, proper diffusion parameters are retrieved, and the determined output color contributes in expectation to underline the structure from which the pixel comes.

Iridescence is the property that makes colors vary by angle of observation. Technology has made it possible to print with ink that has this anisotropic property. The ink that is used contains microparticles and therefore only specific printing methods can be used, for example screen printing. This comes with new demands on halftones, due to its procedure; dots cannot be too small, and keeping structures should be prioritized. To minimize costs and time, prints should be simulated before sending an order. In this thesis, different methods to halftone for screen printing with iridescent ink are developed. Three existing methods are compared to methods that are developed for this specific cause. The result is presented as a 2D-mask and as an interactive 3D-simulation, using data measured from real ink. Also, properties of iridescence are analyzed to separate it from diffuse colors. An OpenGL simulation tool was developed for simulating halftones on 3D-models. The ink reflectance spectra are represented by polynomials that take as input the wavelength and observation angle. Given an observation angle, a spectrum can be found, which can be converted to RGB and set as the output color in the fragment shader. The program uses an RGB-mask which is the combined masks of the halftone. This is loaded as a texture which indicates what polynomial to use. There is no perfect method that works for all types of images. Images that contain colors similar to the inks benefit from morphological halftoning (enhanced structures) for iridescent areas, or foreground, and hatching (an angle dependant grid pattern) for diffuse areas, or background. Other images, that contain mixed colors as cyan, yellow or magenta, benefit froma hue separation error diffusion, where masks are created by thresholding the hue where the mixed colors are error diffused.The actual prints confirmed the hypothesis that tone reproduction can be of lower priority while structure and depth preservance should be highly prioritized. Iridescent colors can be separated from diffuse colors by thresholding the saturation combined with value, or the lightness of the Lab-representation

Dans ce mémoire nous allons présenter une méthode de diffusion d’erreur originale qui peut reconstruire des images en demi-ton qui plaisent à l’œil. Cette méthode préserve des détails fins et des structures visuellement identifiables présentes dans l’image originale. Nous allons tout d’abord présenter et analyser quelques travaux précédents afin de montrer certains problèmes principaux du rendu en demi-ton, et nous allons expliquer pourquoi nous avons décidé d’utiliser un algorithme de diffusion d’erreur pour résoudre ces problèmes. Puis nous allons présenter la méthode proposée qui est conceptuellement simple et efficace. L’image originale est analysée, et son contenu fréquentiel est détecté. Les composantes principales du contenu fréquentiel (la fréquence, l’orientation et le contraste) sont utilisées comme des indices dans un tableau de recherche afin de modifier la méthode de diffusion d’erreur standard. Le tableau de recherche est établi dans un étape de pré-calcul et la modification est composée par la modulation de seuil et la variation des coefficients de diffusion. Ensuite le système en entier est calibré de façon à ce que ces images reconstruites soient visuellement proches d’images originales (des aplats d’intensité constante, des aplats contenant des ondes sinusoïdales avec des fréquences, des orientations et des constrastes différents). Finalement nous allons comparer et analyser des résultats obtenus par la méthode proposée et des travaux précédents, et démontrer que la méthode proposée est capable de reconstruire des images en demi-ton de haute qualité (qui préservent des structures) avec un traitement de temps très faible. / In this work we present an original error-diffusion method which produces visually pleasant halftone images while preserving fine details and visually identifiable structures present in original images. We first present and analyze the previous work to show the major problems in halftoning, and explain why we decided to use an error diffusion algorithm to solve the problems. Then we present our method which is conceptually simple and computationally efficient. The source image is analyzed, and its local frequency content is detected. The main components of the frequency content (main frequency, orientation, and contrast) serve as lookup table indices in a pre-computed database of modifications to a standard error diffusion. The modifications comprise threshold modulation and variation of error-diffusion coefficients. The whole system is calibrated in such a way that the produced halftone images are visually close to original images (patches of constant intensity, patches containing sinu- soidal waves of different frequencies/orientations/contrasts, as well as natural images of different origins). Finally, we compare and analyze the results obtained by our method and previous work, and show that our method can produre high-quality halftone image (which is struc- ture aware) within very short time.