In lossy image compression, the mosquito noise generates near edge when a compression rate is high. We are studying a method to reduce mosquito noise adaptively using the Haar transform with rectangular basis functions. In this paper, we focus on mosquito noise along vertical, horizontal, and diagonal straight edges. Firstly, the 1-dimensional DCT is applied parallel to these edges. Then, we apply the Haar transform to only its DC components. In order to demonstrate the effectiveness of the proposed method, the compression performance is compared with the case where the Haar transform is also applied to the AC components for 400 types of edge images generated by uniform random numbers. Furthermore, numerical experiments using standard images demonstrate subjectively and objectively the superiority of the method over previous methods. The experiment results show the effectiveness of mosquito noise reduction and better PSNR, MSSIM.

Although sign language is one of the most general communication methods used by hearing impaired people, there is a high barrier to communication between impaired people and hearing people, because many of hearing people do not understand sign language. Therefore, if an automatic translation of sign language can be realized, it will contribute to facilitation of communication between them. The authors are now working on realizing sign language translation using an optical camera and a CPU embedded in smartphones as a final goal. In this paper, the authors examine sign language recognition method using the colored glove and the optical camera, and extract six kinds of feature elements for classification from the position of the center of gravity of the colored region of the colored glove and the area of this region. The feature elements are applied to Hidden Markov Model (HMM), Support Vector Machine (SVM), Discriminant Analysis (DA), Linear Classification Model (LCM), k-Nearest Neighbor algorithm (k-NN) and Decision Tree (DT) for classification of each sign language motion. We evaluate the performance of each classifier and propose a method to combine these classification results with the aim of realizing highly accurate recognition. It is shown that recognition performance of 73.1% by a single method, and 76.8% by a combination method, for 35 sign language words can be obtained by the proposed method.

Large-scale digital fabrication is still a challenging issue due to the spatial and material limitations of common 3D printers. In this work, we propose the layered projection mapping approach, an interactive system that helps common users fabricate in large-scale. To verify our system, we utilize balloon art as a case study from its economical and practical aspects. The whole framework is composed of two parts: offline depth calculation and interactive projection guidance. In offline calculation, we first decompose the target 3D model using approximate pyramidal decomposition, then divide the decomposed parts into layers with individual calibrations. In projection guidance, the system provides fabrication guidance with depth differences between the target shape and the current work in real-time progress. Instead of projection of color gradients in previous work, we use the high contrast black and white projection of the numbers in consideration of balloon textures. To increase user immersion, we propose a shaking animation of projected number with significant depth differences. In our case studies, the unskilled participants can build a large-scale balloon art using the proposed system.

Conventionally, visual inspection using image processing was difficult because it was hard to achieve uniform illumination shading in capturing the outer peripheral state of a cylindrical metal product as a camera image. In this paper, verification by experiment in a line camera system that rotates an object 360 degrees and simulation by light reflection characteristics of light are performed. In addition, after clarifying the cause of “unevenness in bright and dark” of the acquired image, the countermeasure method is described.

Paper is made up of fibers with different characteristics based on the raw material used, the conditions in which the paper was stocked, and the process by which it was made. Various studies have been conducted on the structure and formation of paper layers. The fiber of paper used for ancient writings is often analyzed using subjective organoleptic assessment. Furthermore, objective methods to analyze paper stains fibers with chemicals and cause the destruction of the paper. Therefore, depending on the analysis target, nondestructive methods for paper analysis are desired. In this study, we propose an automatic classification of the fibers contained in paper using images obtained with macro photography using a commercially available digital camera with VGG-16. Results show that the average overall accuracy between the predicted classification result and the fiber analysis in three experiments was 94.2%. However, some blurry images taken with a digital camera caused misclassification because determining the structures of fiber layers, such as their thickness and direction, was difficult. In the future, our proposed method could be used to qualify the types of papers in historical and cultural artefacts for example Samarkand paper.