Color Research and Application最新文献

In the present era, hyperspectral images have become a pervasive tool in a multitude of fields. In order to provide a feasible alternative for scenarios where hyperspectral imaging equipment is not accessible, numerous researchers have endeavored to reconstruct hyperspectral information from limited spectral measurements, leading to the development of spectral reconstruction (SR) algorithms that primarily focus on the visible spectrum. In light of the remarkable advancements achieved in many computer vision tasks through the application of deep learning, an increasing number of SR works aim to leverage deeper and wider convolutional neural networks (CNNs) to learn intricate mappings of SR. However, the majority of deep learning methods tend to neglect the design of initial up-sampling when constructing networks. While some methods introduce innovative attention mechanisms, their transferability is limited, impeding further improvement in SR accuracy. To address these issues, we propose a multi-attention interaction and multi-scale fusion network (MAMSN) for SR. It employs a shunt-confluence multi-branch architecture to learn multi-scale information in images. Furthermore, we have devised a separable enhanced up-sampling (SEU) module, situated at the network head, which processes spatial and channel information separately to produce more refined initial up-sampling results. To fully extract features at different scales for visible-spectrum spectral reconstruction, we introduce an adaptive enhanced channel attention (AECA) mechanism and a joint complementary multi-head self-attention (JCMS) mechanism, which are combined into a more powerful feature extraction module, the dual residual double attention block (DRDAB), through a dual residual structure. The experimental results show that the proposed MAMSN network outperforms other SR methods in overall performance, particularly in quantitative metrics and perceptual quality.

Color harmony is an aesthetic sensation evoked by the balanced and coherent arrangement of the colors of visual elements. While traditional methods define harmonious subspaces from geometric relationships or numerical formulas, we employ a data-driven approach to create a unified model for evaluating and generating color combinations of arbitrary sizes. By treating color sequences as linguistic sentences, we construct a color combinations generator using SeqGAN, a generative model capable of learning discrete data through reinforcement learning. The resulting model produces color combinations as much preferred as those by the best models of each size and excels at penalizing color combinations from random sampling. The distribution of the generated colors has more diverse hues than the input data, in contrast to the NLP-based model that predominantly predicts achromatic colors due to exposure bias. The flexible structure of our model allows for simple extension to additional conditions such as group preference or emotional keywords.

Achieving successful cross-media color reproduction is very important in image processing. The purpose of this study is to accumulate high dynamic range data to verify and modify the CAM16-UCS model. There are two experiments in this study. The first experiment was aimed to collect corresponding data of color patches between colors on a display and the real scene viewed under high dynamic range viewing conditions. The results were used to refine CAM16-UCS model. Six illumination levels (i.e., 15, 100, 1000, 3160, 10 000, and 32 000 lx) and 13 test color samples were used in the experiment. Ten observers adjusted the color patches on the display to match the color samples of the real scene. The visual results showed a clear trend, an increase in the illumination level raised vividness perception (both increase in lightness and colorfulness). However, CAM16-UCS did not give accurate prediction to the visual results, especially in the lightness direction. The model was then refined to achieve satisfactory performance and to truthfully reflect the visual phenomena. However, the effect of the modified model could not achieve successful color image reproduction, especially under low illumination conditions. Experiment 2 was conducted by adjusting the overall lightness and colorfulness of the image. The results were used to extend the model for image reproduction. Also, an independent experiment verified that the image generated by the new model matched the real environment well, indicating that the model can perform well in scene restoration.

This paper develops a method based on a stacking ensemble learning model to achieve more accurate conversion from CMYK colors to LAB colors. The model employs tetrahedral interpolation, radial basis function (RBF) interpolation, and KAN as base learners, with linear regression as the meta-learner. Our findings show that the stacking-based model outperforms single models in accuracy for color conversion. In the empirical study, color blocks were printed and the collected data was measured to train and validate the stacking ensemble learning model. The results show that the stacking-based model achieves superior accuracy in color space conversion tasks. This research has substantial practical implications for enhancing color management technology in the printing industry.

This study investigates the ability of artificial intelligence (AI) to simulate human emotional responses to color using two established methods: semantic differential (SD) method and word-color association (WCA) approach. The SD method quantifies emotional reactions to colors through bipolar adjective pairs (e.g., warm–cool, heavy–light), while the WCA method explores associations between specific words and colors. AI responses were compared with data from human participants across various demographics. Results show that AI consistently evaluates basic emotional dimensions, such as warm–cool and heavy–light, with high accuracy, often surpassing human consistency. However, AI struggled with more subjective and culturally influenced dimensions like modern–classical and active-passive. In the WCA experiment, AI replicated many general color associations but faced challenges with complex emotions like joy and anticipation. These findings highlight AI's potential in tasks requiring standardized emotional responses but reveal its limitations in capturing nuanced human emotions, especially in culturally sensitive contexts.

Hans Hass' Under the Caribbean (1954, LI, AT, DE) was one of the world's first underwater colour films. As such, it provides a unique case study and raises interesting questions about the film's colour technology, combining 35 mm chromogenic negative and 16 mm Kodachrome processes with Technicolor imbibition printing in an interweaving of colour processes. Research into the vast amount of Hass' film material held at the Filmarchiv Austria has not yet revealed any of the original Kodachrome footage of this film nor its opticals. However, based on archival documents, it was possible to confirm and reconstruct the workflow Technicolor adopted for this film. Investigating the production history of Under the Caribbean not only provides film historical knowledge of this specific film, but also film technical insights into the production of other films of the early 50s, that also combine several colour processes. This research will be presented together with a discussion of the restoration possibilities offered by the source material, that is, the cut negative and several release prints.