Hand-drawn pictures differ from ordinary perspective images in that the entire scene is composed of local feature regions each of which is projected as seen from its own vista point. This type of projection, called nonperspective projection, has served as one of the common media for our visual communication while its automatic generation process still needs more research. This paper presents a method for automatically generating aesthetic nonperspective images by simulating the deformation principles seen in such hand-drawn pictures. The proposed method first locates the optimal viewpoint for each feature region by maximizing the associated viewpoint entropy value. These optimal viewpoints are then incorporated into the 3D camera parameter field, which is represented by 3D grid samples of camera parameters. Finally, the camera parameter field is smoothed out in order to eliminate any unexpected discontinuities between neighboring feature regions, by taking advantage of image restoration techniques. Several nonperspective images are generated to demonstrate the applicability of the proposed method.

View full abstract

Glare effect is widely used in games and entertainment field as a computer graphics special effect. In this paper, we propose a method to render glare phenomena with high reproduction fidelity and introduce its application to lamp design. Our method simulates a diffraction phenomenon which takes place in the human eye visual system, and pre-computes a glare pattern taking the spectral energy distribution of the light source into account. In rendering, it generates an HDR image from the directional energy distributions of the light source in accordance with the viewing angle, tone-maps and composes the image at the light location. We visualized high intensity light sources such as automobile headlamps directly viewed from the eye and produced results useful to evaluate how glare affects the eyesight.

View full abstract

Wide-angle retinal images are synthesized from multiple fundus photographs (45 degree view) to provide a general view of the fundus. Fundus photographs contain distortion because they represent the projection of a three dimensional curved surfaces on to a two dimensional plane. A technique of making seamless blood vessel joints from several incongruous blood vessels in joint parts is proposed. The evaluation test on our technique and the overlay synthesis technique was executed to 1177 blood vessels. Our technique succeeded in the connection of 307 blood vessels (26%) which were not connected in the overlay synthesis. This method is therefore very promising for the automatic assembly of retinal blood vessel joint processing.

View full abstract

This paper proposes an image search technology that processes character candidate sets by scanning scene images containing character strings such as signboard messages with a recognition dictionary. First, it extracts keywords composed of characters that are regularly aligned in the image by matching the word dictionary to the candidate character set. It then uses the user-input keywords in performing image search. We assume that the principal use of this technology is to retrieve images from data sets held by the user. The proposed technology has the feature that it can extract character strings even if they consist of shaded, three-dimensional characters, lie on a complex background, and are inclined, etc., attributes that cannot be offered by existing character recognition schemes for scene images. To overcome the problem of the background triggering the erroneous recognition of characters, we estimate the transformation angles of the candidate character from angular distributions of the gradient vector of pertinent areas, and add it to the limiting conditions used in determining the arrangement of character strings in the matching against the word dictionary process. Experiments on keyword extraction and image retrieval show that the proposed technology has practical performance in the application assumed, and that its estimation of the transformation angle of candidate characters is effective in decreasing keyword extraction errors.

View full abstract

Polar Harmonic Transform (PHT) is termed to represent a set of transforms those kernels are basic waves and harmonic in nature. PHTs consist of Polar Complex Exponential Transform (PCET), Polar Cosine Transform (PCT) and Polar Sine Transform (PST). They are proposed to represent invariant image patterns for two dimensional image retrieval and pattern recognition tasks. They are demonstrated to show superiorities comparing with other methods on describing rotation invariant patterns for images. Kernel computation of PHTs is also simple and has no numerical stability issue. However in order to increase the computation speed, fast computation method is needed especially for real world applications like limited computing environments, large image databases and realtime systems. This paper presents Fast Polar Harmonic Transforms (FPHTs) including Fast Polar Complex Exponential Transform (FPCET), Fast Polar Cosine Transform (FPCT) and Fast Polar Sine Transform (FPST) that are deduced based on mathematical properties of trigonometric functions and number theory. The proposed FPHTs are averagely over 10 times faster than PHTs that significantly boost computation process. The experimental results on both synthetic and real data are given to illustrate the effectiveness of the proposed fast transforms.

View full abstract

Estimating salient regions of an image has a very important role in areas such as image compression, evaluation, searching, robot vision and so on. In this case, the most difficult problem is to estimate those regions without any prior knowledge. In this paper, we present an estimating method of salient regions in color natural images based on the stability of local extrema in scale-space. When the object's region has more stable structure compared to other area, it must be more salient. So the saliency of the regions must be defined based on its stability in scale-space. In our method, local extrema of images are considered describing the complexity of objects and the background in the image. Salient regions are estimated based on the stability of the local extrema for the blurring parameter. Comparing the human map of fixations of the eye-movement recording, we confirm that our method successfully estimates the salient regions.

View full abstract

This paper presented a novel digital refocus method that can render blur effects with less ghosting artifacts even from under-sampled light fields. The presented method generates a set of images focused on multiple depths by using the conventional synthetic aperture method with larger aperture size and then converts it to a set of ghost-suppressed images with the desired aperture size. This paper showed that this conversion can be achieved by a 3D filtering in the frequency domain and that the filter was derived based only on multi-camera settings independent of scene information. Effectiveness of the presented method was valied through experiments using both synthetically and real images.

View full abstract

In traditional image composition methods for cutting out an object (source object) from an image (source image) and pasting it onto another image (target image), users have to segment a foreground object (target object) in a target image or cut out the part of the source object when they want to partially hide a source object behind it. While recent image editing tools greatly facilitate segmentation operations, it can be tedious to segment each object if users try to place a source object in various positions in a target image before satisfying. We propose a method which allows users to drag a source object and slip it behind a target object, so that users can move a source object around without manually segmenting each part of a target image.

View full abstract

This paper describes about the media which touch off informal text communication by black spot stimulus. A black spot stimulus is shown as a background image of the text field in advance of text input. The purpose of black spot stimulus presentation is to touch off conceiving topic based on the cognitive process model of creative thinking. The prototype system based on this media concept was developed, and the communication experiment was conducted. It turned out as a result of the experiment that these media touched off the idea about making topic strongly in the text communication. Moreover, it was implied that the topic touched off deepens the mutual understanding between speakers.

View full abstract

In this paper, a new design scheme of multi-dimensional linear-phase lapped orthogonal transform (LOT) for image and audio coding is proposed. LOT basis functions described by (N×2N) matrix can be generated by rotation model of Nth-order normalized orthogonal vectors in 2N-dimensional orthogonal space. By using this model, the minimum numbers of parameters which optimize coding gain were already reported. We examined the relation between rotational parameters and coding gain, and investigated the sequence of rotational operations, which distribute the power of LOT bases from center to edge. By eliminating the parameters which have low value and low sensitivity, substantial number of parameters can be reduced without the loss of coding gain. Furthermore several design examples of LOT are given to validate the proposed scheme.

View full abstract

We propose a new global binarization method suited for character strings extraction from a scene image. To obtain a high-quality segmentation result, binarization is a significant step in character extraction process. Since character strings region of sign board has the wide variety of layout and color, it is hard to binarize this category of images by conventional global methods, which are based on statistical information. The proposed method can select a threshold by evaluating the fineness of the binarized image by using the fractal dimension. By evaluating the compactness and stability of the regions of binarized image with the fractal dimension calculated by blanket method, the proposed method achieves fine binarzation results, which are no way inferior to manually and/or using one of promising global methods known as Otsu binarizing method.

View full abstract

We have been investigating the continuous tone recording and erasing characteristics using the reversible thermo sensitive material comprising of a resin and organic low molecular weight material and so on. The principal results are as follows. To begin with, we examined the pulse number control process from non-stable heat transfer function, and then clarified the optimum recording conditions, specified with the characteristics of applied energy, recording speed, and thermal response. Next, with thermal control, the imaging data can be kept stable even when the temperature of the thermal head varies. Finally, we found out that the object-oriented architecture of recording engine is effective from the viewpoints of energy-saving, high-level image quality reproduction and cost reduction.

View full abstract

This paper proposes a system that automatically deforms the original video images into line drawing expression transmitted in both directions so as to achieve two goals: easing the concerns of users privacy and ensuring good enough support by visual channel. We examine the effect of line drawing video for support system on the effectiveness and impression of privacy on a collaboration between a novice user and an operator. As a result, it became clear that concern of the user on privacy or security are greatly reduced by using a line drawing image, and there was no significant difference on efficiency with an original image and a line drawing image.

View full abstract

The Hilditch thinning algorithm is a sophisticated method for digital images and thought to be still useful in today. A kind of faulty version of Hilditch thinning algorithm which acts on incorrect connectivity condition has been found in some theses and/or on web sites. In this paper, we point out why and how the false version has appeared to request attention to algorithm details to use a strict Hilditch thinning algorithm.

View full abstract