Search

The process of adapting or repurposing manga pages is a time-consuming task that requires manga artists to manually work on every single screentone region and apply new patterns to create novel screentones across multiple panels. To address this issue, we propose an automatic manga rescreening pipeline that aims to minimize the human effort involved in manga adaptation. Our pipeline automatically recognizes screentone regions and generates novel screentones with newly specified characteristics (e.g., intensity or type). Existing manga generation methods have limitations in understanding and synthesizing complex tone- or intensity-varying regions. To overcome these limitations, we propose a novel interpretable representation of screentones that disentangles their intensity and type features, enabling better recognition and synthesis of screentones. This interpretable screentone representation reduces ambiguity in recognizing intensity-varying regions and provides fine-grained controls during screentone synthesis by decoupling and anchoring the type or the intensity feature. Our proposed method is demonstrated to be effective and convenient through various experiments, showcasing the superiority of the newly proposed pipeline with the interpretable screentone representations., 10 pages, 11 figures

Video colorization is a challenging task that involves inferring plausible and temporally consistent colors for grayscale frames. In this paper, we present ColorDiffuser, an adaptation of a pre-trained text-to-image latent diffusion model for video colorization. With the proposed adapter-based approach, we repropose the pre-trained text-to-image model to accept input grayscale video frames, with the optional text description, for video colorization. To enhance the temporal coherence and maintain the vividness of colorization across frames, we propose two novel techniques: the Color Propagation Attention and Alternated Sampling Strategy. Color Propagation Attention enables the model to refine its colorization decision based on a reference latent frame, while Alternated Sampling Strategy captures spatiotemporal dependencies by using the next and previous adjacent latent frames alternatively as reference during the generative diffusion sampling steps. This encourages bidirectional color information propagation between adjacent video frames, leading to improved color consistency across frames. We conduct extensive experiments on benchmark datasets, and the results demonstrate the effectiveness of our proposed framework. The evaluations show that ColorDiffuser achieves state-of-the-art performance in video colorization, surpassing existing methods in terms of color fidelity, temporal consistency, and visual quality., project page: https://colordiffuser.github.io/

Image colorization has been attracting the research interests of the community for decades. However, existing methods still struggle to provide satisfactory colorized results given grayscale images due to a lack of human-like global understanding of colors. Recently, large-scale Text-to-Image (T2I) models have been exploited to transfer the semantic information from the text prompts to the image domain, where text provides a global control for semantic objects in the image. In this work, we introduce a colorization model piggybacking on the existing powerful T2I diffusion model. Our key idea is to exploit the color prior knowledge in the pre-trained T2I diffusion model for realistic and diverse colorization. A diffusion guider is designed to incorporate the pre-trained weights of the latent diffusion model to output a latent color prior that conforms to the visual semantics of the grayscale input. A lightness-aware VQVAE will then generate the colorized result with pixel-perfect alignment to the given grayscale image. Our model can also achieve conditional colorization with additional inputs (e.g. user hints and texts). Extensive experiments show that our method achieves state-of-the-art performance in terms of perceptual quality., project page: https://piggyback-color.github.io/

Microfluidics-based fabrication of cell-laden microgels has shown great potential for applications in cell therapy and tissue engineering, however, the difficulty in chip operation and compromised cell viability due to cell sedimentation and channel blockage remain a major challenge for functional cell-laden microgels preparation. Herein, we presented the design and optimization of integrated microfluidic chip for large-scale preparation of cell-laden microgels with controllable size and complex microstructure. Specifically, to avoid severe cell sedimentation and uneven distribution in the parallelized microchannel, we simulated cell movement state using computational fluid dynamics simulation. It was found that higher laminar flow velocity gradient and higher precursor viscosity can significantly improve the uniform cell distribution in parallelized channels and reduce the product difference between channels. Moreover, we designed multiple-layered microfluidic chips allowing multiple inputting liquids for the fabrication of microgels with complex structures. This integrated chip facilitated cell encapsulation at a maximum production rate of 240 ml/h of cell suspension with retained cell viability and functionality. Therefore, our study provided a biocompatible and high-throughput strategy for large-scale preparation of cell-laden microgels, which can enable significant advances for clinical-relevant applications of cell-laden microgels, including cell therapy, tissue regeneration and 3D bioprinting.

Although digital cameras can acquire high-dynamic range (HDR) images, the captured HDR information are mostly quantized to low-dynamic range (LDR) images for display compatibility and compact storage. In this paper, we propose an invertible tone mapping method that converts the multi-exposure HDR to a true LDR (8-bit per color channel) and reserves the capability to accurately restore the original HDR from this {\em invertible LDR}. Our invertible LDR can mimic the appearance of a user-selected tone mapping style. It can be shared over any existing social network platforms that may re-encode or format-convert the uploaded images, without much hurting the accuracy of the restored HDR counterpart. To achieve this, we regard the tone mapping and the restoration as coupled processes, and formulate them as an encoding-and-decoding problem through convolutional neural networks. Particularly, our model supports pluggable style modulators, each of which bakes a specific tone mapping style, and thus favors the application flexibility. Our method is evaluated with a rich variety of HDR images and multiple tone mapping operators, which shows the superiority over relevant state-of-the-art methods. Also, we conduct ablation study to justify our design and discuss the robustness and generality toward real applications.

To share the same visual content between color vision deficiencies (CVD) and normal-vision people, attempts have been made to allocate the two visual experiences of a binocular display (wearing and not wearing glasses) to CVD and normal-vision audiences. However, existing approaches only work for still images. Although state-of-the-art temporal filtering techniques can be applied to smooth the per-frame generated content, they may fail to maintain the multiple binocular constraints needed in our applications, and even worse, sometimes introduce color inconsistency (same color regions map to different colors). In this paper, we propose to train a neural network to predict the temporal coherent polynomial coefficients in the domain of global color decomposition. This indirect formulation solves the color inconsistency problem. Our key challenge is to design a neural network to predict the temporal coherent coefficients, while maintaining all required binocular constraints. Our method is evaluated on various videos and all metrics confirm that it outperforms all existing solutions.

Quantitative gene expression analysis is indispensable for accurately characterizing the expression levels of genes involved in the response and acclimation of algae to environmental stresses. However, appropriate reference genes in the genus Bangia (order Bangiales) for quantification of stress-inducible gene expression are not well described. Thus, we evaluated the suitability of genes encoding 18S ribosomal RNA (18S rRNA), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), actin, tubulin, elongation factor 1 (EF1), and 60S ribosomal RNA (60S rRNA) as reference genes for real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR) under various stress conditions in ‘Bangia’ sp. ESS1, classified into the ‘Bangia’ 2 group. Our results showed that the 60S rRNA and GAPDH genes were most suitable for gene expression analyses in response to nutrient deficiency, high salinity, and temperature stress, whereas normalization of desiccation-inducible expression required use of the EF1 and tubulin genes. These findings indicate that reference genes for analysis of stress-inducible gene expression by qRT-PCR should be selected according to the environmental stress being studied. Our efforts to identify easily amplified reference genes for various stress responses provide novel insights into the selection of reference genes for seaweeds.

An equivalent quantization correction technique for the pipelined successive approximation register (PI-SAR) analog-to-digital converter (ADC) is proposed. The technique corrects both the capacitance mismatch and the residual amplifier (RA) gain errors. It utilizes the low capacitors to quantify the digital weight of each capacitor, and corrects the RA gain errors by the digital weight of capacitors. Based on a 14-bit 100MSs PI-SAR ADC with the calibration technique, the validity of the technique is fully verified. The implementation of the correction technique requires only simple control logic of analog circuit and adders along with multiplexers of digital circuit. The simulation results indicate that the proposed calibration technique enhances the performance of the ADC effectively. SFDR, SNDR improved from 53.33dB, 47.33dB to 102.21dB, 82.20dB, respectively. The digital circuit only contains 6162 gates and consumes 0.154mW.

この博士論文全文の閲覧方法については、以下のサイトをご参照ください。, https://www.lib.hokudai.ac.jp/dissertations/copy-guides/, (主査) 教授 岸村 栄毅, 教授 細川 雅史, 准教授 井上 晶, 准教授 三上 浩司, 水産科学院（海洋応用生命科学専攻）

(主査) 教授 岸村 栄毅, 教授 細川 雅史, 准教授 井上 晶, 准教授 三上 浩司, 水産科学院（海洋応用生命科学専攻）

With the rapid development of mobile Internet, mobile Internet has come into the era of big data. The demand for data analysis of mobile applications has become more and more obvious, which puts forward higher requirements for the standard of mobile application information collection. Due to the large number of applications, almost all third-party app stores display only a small number of applications, and most of the information is hidden in the Deep Web database behind the query form. The existing crawler strategy cannot meet the demand. In order to solve the above problems, this paper proposes a collection method based on category keywords query to improve the crawl rate and integrity of the mobile app stores information collection. Firstly, get the information of application interfaces that include various kinds of applications by using the vertical crawler. Then extract the keywords that represent each category of applications by TF-IDF algorithm from the application name and description information. Finally, incremental crawling is performed by using keyword query-based acquisition method. Results show that this collection method effectively promoted information integrity and acquisition efficiency.

A real-time temperature detection circuit with 0.1°C accuracy is presented. The temperature sensor is composed by a current controlled oscillator (ICO). Its oscillating frequency is proportional to the temperature but independent with the power supply. The frequency change will be digitised by a counter and thus realise the temperature detection. The detection accuracy is limited by the linearity of ICO. In order to improve temperature detection accuracy, a temperature segmentation algorithm is proposed. The temperature sensor is implemented with 0.18 μm CMOS technology and the core area is 0.023 mm2. The power consumption of the proposed sensor is 0.3 mW with 1.8 V supply. Experimental results show that this sensor can detect the temperature from −40 to 125°C with 0.1°C accuracy and the detection error is within ±0.8°C.

Security problem of mobile application is a great challenge that mobile internet facing nowadays. Traditional detection and protection methods can't protect users from the emergence of a large number of malwares well, nor can they prevent normal applications from being decompiled and repacked by hackers. In that case, this paper proposes a security model of mobile application which includes two main stages. First, static and dynamic analysis technology are combined together and a double-staining algorithm is proposed to catch application behaviours so as to improve the accuracy of analysis. Then a reinforce framework for mobile application is proposed based on encryption and custom loader. The proposed model can not only detect whether applications are malicious, but also protect apps from being tampered by malicious developers, which will promote the development of security system of mobile internet.

We utilize the angle and momentum resolution of our apparatus to study the polarization of 375 GeV/c Sigma(+) hyperons produced by 800 GeV/c protons incident on a Cu target. By examining in detail two of our high statistics data samples, we find evidence for structure in the p(t) dependence of Sigma(+) polarization and are able to extract the x(F) dependence of the Sigma(+) polarization and compare it with x(F) behavior in the Lambda(0) and Xi(-) systems.

We have measured the branching ratio for the hyperon radiative decay ${\mathrm{\ensuremath{\Sigma}}}^{+}$\ensuremath{\rightarrow}p\ensuremath{\gamma} using the Fermilab polarized charged hyperon beam. This measurement and our previously published result on the asymmetry parameter in the same decay are part of Fermilab experiment E761. We find B(${\mathrm{\ensuremath{\Sigma}}}^{+}$\ensuremath{\rightarrow}p\ensuremath{\gamma})/B(${\mathrm{\ensuremath{\Sigma}}}^{+}$\ensuremath{\rightarrow}p${\mathrm{\ensuremath{\pi}}}^{0}$) to be [2.32\ifmmode\pm\else\textpm\fi{}0.11(stat)\ifmmode\pm\else\textpm\fi{}0.10(syst)]\ifmmode\times\else\texttimes\fi{}${10}^{\mathrm{\ensuremath{-}}3}$ with a sample of 31 901 events. The higher statistics and careful attention to systematic uncertainties make these significant improvements over previous measurements. We describe how our measurements were performed and briefly review the theoretical implications of these results.

We have measured the branching ratio for the hyperon radiative decay [Xi][sup [minus]][r arrow][Sigma][sup [minus]][gamma] from a sample of 211[plus minus]33 events obtained in the polarized 375 GeV/[ital c] charged hyperon beam at Fermilab. We find [ital B]([Xi][sup [minus]][r arrow][Sigma][sup [minus]][gamma]/[Xi][sup [minus]][r arrow][Lambda][sup 0][pi][sup [minus]])=(1.22[plus minus]0.23[plus minus]0.06)[times]10[sup [minus]4] where the quoted errors are statistical and systematic, respectively. We have also obtained an indication that the sign of the asymmetry parameter of this decay is positive.

We have measured the magnetic moments of the SIGMA+ and SIGMABAR- hyperons produced by 800-GeV protons incident on a Cu target. We determine the SIGMA+ magnetic moment to be (2.4613 +/- 0.0034 +/- 0.0040)mu(N) where the uncertainties are statistical and systematic, respectively. In this first measurement we determine the magnetic moment of the SIGMABAR- to be -(2.428 +/- 0.036 +/- 0.007)mu(N). The magnetic moments of the SIGMA+ and SIGMABAR- are consistent with each other in magnitude but opposite in sign as required by CPT invariance.

We have measured the polarization of 375-GeV/c SIGMA+ and SIGMABAR- hyperons produced by 800-GeV/c protons incident on a Cu target. We find that the SIGMA+ polarization rises with increasing p(t) to a maximum of 16% at p(t) = 1.0 GeV/c and then decreases to 10% at p(t) = 1.8 GeV/c. We compare this SIGMA+ polarization with data at lower energies. The SIGMABAR- polarization has been measured for the first time. It has the same sign as the SIGMA+ but smaller magnitude in a similar kinematical region.

We have measured the asymmetry parameter (${\mathrm{\ensuremath{\alpha}}}_{\ensuremath{\gamma}}$) in the hyperon radiative decay ${\mathrm{\ensuremath{\Sigma}}}^{+}$\ensuremath{\rightarrow}p\ensuremath{\gamma} with a sample of 34 754\ifmmode\pm\else\textpm\fi{}212 events obtained in a polarized charged hyperon beam experiment at Fermilab. We find ${\mathrm{\ensuremath{\alpha}}}_{\ensuremath{\gamma}}$=-0.720\ifmmode\pm\else\textpm\fi{}0.086\ifmmode\pm\else\textpm\fi{}0.045, where the quoted errors are statistical and systematic, respectively.

We have searched for the rare hyperon radiative decay Omega- to Xi- gamma, using the Fermilab Proton Center 375 GeV/c charged hyperon beam. Our measurement of the Omega- beam fraction at 13.6 m from production is (4.8 +_ 0.4)10**(-5). No signal was found and we determine a new upper limit of (4.6 x 10**-4) at 90% CL for the Omega- to Xi- gamma branching ratio.

The design of a 1.5 m long NdFeB permanent magnet undulator is described. The characteristics are presented. It consists of 400 blocks of permanent magnets (7.5 × 7.5 × 40 mm3) arranged in a Halbach configuration and provides a periodic magnetic field with a peak field of 0.45 T for a gap of 1 cm.