Recently, augmented reality (AR) by using a projector, which is often called projection-based AR, has been focused on and widely studied. We propose a projection-based AR system, named “Lumipen” for a high-speed or high-frequency object, using a 1000-fps vision sensor and a high-speed optical axis controller. Conventional research on projection-based AR has mainly focused on static objects or circumstances. In our system, the projected direction and projected image can be controlled based on the dynamic object behavior detected by a 1000-fps vision sensor. Instead of a rotational basement, the high-speed optical axis controller is an essential subsystem to catch up with the speed of the 1000-fps vision sensor. Lumipen is expected to provide high-quality visual information addition even in dynamic live content (e.g. a play, sport game, music concert) similar to CG in video content.
This research investigates the effect of viewing conditions on depth perception under the conditions that the same stereoscopic images are presented on different stereoscopic displays. Such conditions often occur in practical situations. We measured the differential limens and the point of subjective equality Using three common types of stereoscopic displays in psychophysical experiments. The experimental results showed two properties of depth perception based on the differences among the stereoscopic displays. One is the tendency that the depth is perceived to be closer than the depth presentation position when the position is far away, i.e., at a distance over roughly 800 cm. The other has been obtained from analysis of variance to be roughly 15 min in the disparity angle of the stereoscopic images as the differential limens, which is in the vicinity of the points of subjective equality.
This paper proposes a detection method of artificial obstacles from a single image. The proposed method first detects local lines as candidates based on the duality of gradient orientation and then globally determines whether they are artificial obstacles by using fractal dimension and the perimeter of the region. The proposed method has two distinct advantages: (1) the accurate detection of artificial obstacles in the form of not only straight lines but also curved lines, and (2) the fast detection of obstacles by using hysteresis thresholding at the global detection phase. Results with real images are presented to illustrate the superior performance of the proposed method.
In this paper, we propose a method of model-based integral 3D contents production from multi-view images to produce integral 3D contents of sports scenes in TV programs. In this method, we use a depth estimation method with belief propagation to create 3D models at each view point of multi-view images and convert these models into an integral 3D image. The remarkable feature of our method is that it uses the energy distribution kurtosis of belief propagation to inhibit the effects of depth estimation errors. We estimate the reliability of the depth estimation results by the energy distribution kurtosis and create an integral 3D image that inhibits the effect of depth estimation errors by integrating 3D model parts with high reliability. In experiments, we focus on sumo wrestling scenes in a TV program and create unprecedented integral 3D contents of the sports scenes.
There are two important issues for accurate concept detection in videos. One is to train a concept detector with a large number of training examples. The other is to extract the feature representation of a shot based on descriptors, which are densely sampled in both the spatial and temporal dimensions. This paper describes two fast and exact methods based on matrix operation, where a large amount of data are processed in a batch without any approximation. The first method trains a concept detector based on batch computation of similarities among many training examples. The second method extracts the feature representation of a shot by computing probability densities of many descriptors in a batch. The experimental results validate the efficiency and effectiveness of our methods. In particular, the concept detection result obtained by our methods was ranked top in the annual worldwide competition, TRECVID 2012 Semantic Indexing (light).
In this paper, DoA(direction of arrival) in a measurement system that estimates the conditions of propagation paths for digital terrestrial television broadcasting is used. The broadcasting signals are received by two different antennas that have different roles. One is used for receiving the reference signal to obtain the synchronization timing. The other antenna is rotated in a horizontal surface to estimate the DoA. By analyzing of the received datas, a phase compensation technique and delay profile synchronization technique are proposed. The delay profiles that are measured against each measurement angle are synchronized to the reference delay profile using the evaluation function and Newton method. The evaluation function is defined as the square error of reference and target delay profile. The directions of arrival of the paths are estimated from the synchronized delay profiles. It is possible to synchronize delay profiles using the proposed scheme.
We compared the visual fatigue experienced during viewing on a 2D and a 3D television. Twenty participants were individually shown an 80-min clip of the motion picture Avatar on a 2D and a 3D television. Before and after each viewing, the participants completed a questionnaire assessing visual fatigue and discomfort. The visual acuity, accommodation speed and heterophoria of each participant were measured on another day. The results show that the 3D television caused greater visual fatigue than when the 2D television was viewed. Substantial individual differences in visual fatigue were found between participants in the 3D case and these differences varied with the accommodation speed for the non-dominant eye, the difference in visual acuity between eyes and the amount of horizontal heterophoria. Relationships between visual functions and visual fatigue during movie viewing on 3D televisions are discussed.
A severe spectrum shortage for wireless communications poses spectrum overlapping and interference problems with a significant degradation in performance. To reduce the number of errors, using techniques to mitigate interference is essential. We investigate techniques using Viterbi decoding combined with QPSK in terms of the bit error rate obtained by both computer simulations and upper bounds derived by the moments method. Analytical results show the superiority of metric limiting, which yields a lower bit error rate performance against interferences.
Many methods that construct image classification algorithms automatically using evolutionary computation have been studied. Although these classifiers are very effective, several problems have been pointed out. For example, it is difficult to analyze or modify classifiers, and they are too complicated for humans to understand. In this paper, we propose a new method for classification using an evolutionary decision network (EDEN) that emphasizes good human-readability. EDEN automatically constructs an adequate network for classification by combining simple nodes using evolutionary computation. This network is composed of a set of nodes that changes the branches of the decision flow in accordance with the feature values of the input data. We build an effective classifier by optimizing a set of nodes and their threshold values for branching. In experiments, we evaluate EDEN by applying it to image classification problems. The experimental results show EDEN is able to build an effective classifier with a human-readable structure and achieve satisfactory performance.