Search

We report a novel approach for achieving efficient zooming/panning without resolution loss in streaming videos for mobile devices such as smartphones and tablets. It is in principle possible to zoom without losing video quality by playing videos with significantly higher resolutions than the display resolution. However, this can drastically increase network traffic and the processing burden of terminal devices. To address this problem, we propose an efficient distribution of a region of interest in video frames with the appropriate resolution based on the zoom factor. We implement the proposed architecture as a web application based on HTTP adaptive streaming and demonstrate its feasibility.

Compressive spectral imaging (CSI) is an emerging technique for acquiring hyperspectral images based on compressed sensing theory. By combining powerful data-driven methods with traditional iterative convex optimization algorithms, we propose a multi-staged neural network architecture, which is inspired by ADMM, to achieve fast and accurate reconstruction for CSI. All parameters of ADMM and deep hyperspectral priors are jointly learned. The proposed ADMM-inspired network was experimentally shown to outperform an existing state-of-the-art reconstruction method in terms of both speed and accuracy.

This paper shows a STT (State Transition Tuple) coverage criterion for extended place/transition net-based testing. A STT as a measuring object consists of state transitions whose execution may trigger a target failure. The test efficiency is evaluated based on the ratio of STTs executed by test cases.

Neural word vector (NWV) such as word2vec is a powerful text representation tool that can encode extensive semantic information into compact vectors. This ability poses an interesting question in relation to image processing research - Can we learn better semantic image features from NWVs? We empirically explore this question in the context of semantic content-based image retrieval (CBIR). In this paper, we consider cross-modal transfer learning (CMT) to improve initial convolutional neural network (CNN) image features by using NWVs. We first show that NWVs can improve semantic CBIR performance compared to classical word vectors, even if it is with simple CMT models, i.e., canonical correlation analysis (CCA). Next, inspired by a characteristic property of NWVs, we propose a new CMT model and demonstrate that it can improve CBIR performance even further.

We propose a novel method for the recognition of objects that match a given 3D model in large-scale scene point clouds captured in indoor environments with a laser range finder. Since large-scale indoor point clouds are greatly damaged by noise such as clutter, occlusion, hole, and measurement errors, it is difficult to exactly identify local correspondences between points in a target model point cloud and points in a scene point cloud, based on similarities between local descriptors computed at keypoints on both point clouds. To avoid such a problem, we suggest to utilize sliding window in order to match the input model and pieces of scene point clouds, both of which are represented with Bag-of-Features(BoF). A BoF representation of a window is efficiently calculated by using the integral image, which stores accumulated BoF vectors. Though BoF is robust to partial noises, it does not preserve any spatial information. Then, we propose a method to make a global descriptor of a window which is almost invariant to horizontal rotations of an object inside the divided window and roughly preserves spatial information by dividing sliding window into several parts. Experiments on real world data show that our approach offers better performance than a baseline method in terms of precision and recall.

We tackle the problem of joint discovery and segmentation of the object of interest from noisy image sets collected via web crawling (e.g., Figure 1). Existing methods [1] [2] [3] employ region-wise comparison in order to separate noise images (images not containing target objects) from the rest, which may be a bottleneck for scaling up to larger datasets. Our idea to avoid such computationally intensive operations is to use image-wise reconstruction errors. Specifically, based on the assumption that images containing target objects are easier to be reconstructed by a pool of target objects than noise images, we first reconstruct each image using a small number of similar target objects. The resulting error is then combined with some other criteria (e.g., saliency) so as to delineate only target object regions. Experimental evaluations on a noisy image dataset [1] demonstrate that our approach achieves state-of-the-art results on every subset of the dataset 5–7 times faster than existing methods.

The purpose of this study is to verify our live video transmission specifications in a real environment. We implemented a high-definition video transmitter that adjusts the video encoding rate according to the physical link status using a cross-layer rate control scheme. The transmitter will be used as reference equipment to practically evaluate the many simulated algorithms, because rate-control algorithms can be implemented in software regardless of the kind of physical layer. First, we explain the transmitter architecture, and then, report on our evaluation of the effectiveness of our cross-layer rate-control algorithm in comparison with a single application layer rate-control algorithm. We verified our algorithm can more quickly and accurately follow the link status than the single application layer rate-control algorithm using the implemented transmitter. Furthermore, we introduce an example in which the equipment can be easily used for other physical links because the reference equipment architecture abstracts the physical layer.

In this paper, we describe an approach to accelerate fingerprint techniques by skipping the search for irrelevant sections of the signal and demonstrate its application to the divide and locate (DAL) audio fingerprint method. The search result for the applied method, DAL3, is the same as that of DAL mathematically. Experimental results show that DAL3 can reduce the computational cost of DAL to approximately 25% for the task of music signal retrieval.

Crowd analysis using cameras has attracted much attention for public safety and marketing. Among techniques of the crowd analysis, we focus on spatial people density estimation which estimates the number of people for each small area in a floor region. However, spatial people density cannot be estimated accurately for an area far from the camera because of the occlusion by people in a closer area. Therefore, we propose a method using a memory based regression method with images captured from cameras from multiple viewpoints. This method is realized by looking up a table that consists of correspondences between people density maps and crowd appearances. Since the crowd appearances include situations where various occlusions occur, an estimation robust to occlusion should be realized. In an experiment, we examined the effectiveness of the proposed method.

This paper proposes a video retrieval method that detects frame sections that correspond to shots in a query (video segment) with single frame level accuracy. The method adopts the coarse-to-fine strategy to decrease the processing time and the memory consumption, dynamic threshold with initial ranges for small segments is proposed to detect the exact beginning and end of each corresponding frame section to each shot in a query. Experiments on real videos show that our method can achieve accurate video detection with exact frame position while reducing processing time and memory consumption.

Due to the increasing number of obese people, this paper presents a Fuzzy Markup Language (FML)-based recommender system for restaurants to infer the recommended level of a restaurant by giving the estimated calories in food, the distance to the restaurant, and the price of the served food. With the standard of the Food Exchange List (FEL), developed by The Japan Diabetes Society, and the information about the food calories, we can estimate the calories of different kinds of the food. We first construct the ontology of this system, and then we use the FML to build the knowledge base and the rule base of the recommender system based on domain experts' suggestions and Japanese persons' experience in diet. With the input of the estimated calories of food, distance of the restaurant, and price of the food, the recommended level of the restaurant is inferred to allow people to both enjoy the delicious food and make an exercise.

This paper presents a Fuzzy Markup Language (FML)-based Japanese diet assessment system and applies it to Japanese diet. Food Exchange List (FEL), published by Japan Diabetes Society, is adopted as a standard to assess one person's diet healthy level to help him/her eat a wide variety of foods and drinks to balance the eaten food items and physical activity. FML is used to describe the knowledge base and rule base of the diet domain. The dietician first defines the nutrient facts of the collected food from food guidebook and Internet. Then, the involved subject records his/her personal information and daily meals for a constant period. Based on the predefined Japanese food ontology, including ingredients of each food item and the contained units of each food group, the proposed fuzzy inference mechanism is implemented to infer the possibility of dietary healthy level for one-day meal. From the simulation results, the proposed approach is feasible to apply to Japanese diet assessment.

With the advance of an aging society, people who are physically handicapped have specific needs concerning mobility assistance in relation to their respective living conditions. Moreover, operating an electric wheelchair indoors in confined spaces requires considerable skill. This paper presents an obstacle avoidance support system for an electric wheelchair, using reinforcement learning. The obstacle avoidance is semi-automatically supported by the Minimum Vector Field Histogram (MVFH) method. The MVFH modifies the user manipulation and assists the obstacle avoidance. In the proposed scheme, the modification rate is adjusted by reinforcement learning according to the environment and the user condition. The newly proposed scheme is numerically evaluated on a simulation example. Furthermore, the proposed scheme was applied to an experimental electric wheelchair, and the effectiveness of the proposed technique was verified in a real operating environment.

A flexible and transparent loudspeaker driven by PVDF (polyvinylidenefluoride) is proposed. Fabricated prototype loudspeaker had flexibility and transparency, and its acoustic characteristics and motion principle were investigated. The investigation yielded low distortion characteristics of the proposed loudspeaker, and such advantages originated from the assemblage of the plastic film and the PVDF.

The super hi-vision (SHV) developed by NHK is an ultrahigh-definition video system with 4000 scanning lines. Its video format comprises 7680times4320 pixels, which is 16 times the total number of pixels of HDTV (high definition television) and the frame rate is 60 Hz with progressive scanning. It has been designed to give viewers a strong sense of reality. In order to make the system suitable for practical use such as broadcasting services, a high-efficiency compression coding system is necessary. Therefore, we have developed several super hi-vision codec systems based on MPEG-2 and AVC/H.264 video coding standards. SHV video signals are converted to sixteen HDTV signals and each video signal is processed by HDTV codec. The codec systems share some characteristics such as the frame synchronized encoding function. In this paper, details of these codec systems are described and transmission experiments using the codec are introduced.

In this paper, a postural supporting device using a balloon actuator-sensor system (BASS) is proposed for the person with disability who is unable to move their body because of injury or disease. The BASS is able to control the position and stiffness adaptively using an adaptive learning impedance controller. The pneumatic actuator has excellent compliance and flexibility, therefore that is good for the human-mechanical system. However, it also has nonlinearity, hence high precision control is difficult. Therefore, in this paper, the CMAC-PID control scheme is installed. The control performance of the BASS is evaluated by the control experiment.

Signal similarity search is an important technique for music information retrieval. A basic task is finding identical signal segments on unlabeled music-signal archives, given a short music signal fragment as a query. In such a task, the search must be fast and sufficiently robust against possible signal fluctuations due to noise and distortions. In this special session paper, we describe a search method designed to cope with additive interfering sounds by spectral partitioning. Then, we introduce another method designed to be robust under multiplicative noise or distortion based on binary area representation.

This paper proposes training equipment using a balloon actuator-sensor system (BASS) for persons unable to move their hands because of injury or disease. BASS is able to control the stiffness adaptively using an adaptive learning impedance controller. The pneumatic actuator has excellent compliance and flexibility, which is good for a human-mechanical system. However, it is also nonlinear, hence high precision control is difficult. Therefore, a CMAC-PID control scheme is installed. Finally, the BASS control performance is evaluated in a control experiment.

In this paper, we demonstrate a novel poster image matching system for wireless multimedia applications. We propose a method that incorporates both color and layout information of the poster image to achieve a robust performance in poster image matching. We apply both color compensation and background separation to extract a poster from an image effectively. Based on our experiment, we show that even under the effects of lighting, image rotation, scaling, and occlusion, our system can still maintain high recall and precision. We also show that our system can recognize the correct image from a database which contains several poster images with similar features. Finally, the promising performance of the poster image matching encourages us to further enrich the information retrieval for wireless environment.

This paper describes a prototype system of an Mm (mega-pixel and milli-second) vision camera head, which strikes a balance between high resolution and high speed. This system is designed based on the concept of MmVision with intelligent pixel selection; target tracking results at a 1000 fps level are shown.

In this paper, a movable neck supporter (MNS) is proposed for the handicapped persons who are unable to move their necks from his own motive with a high degree of disability. The proposed system has a parallel link mechanism using several pneumatic actuators. It assists the twisting and nodding motion of the user's head. By the way, the head motion is different individually, and it is always changing by his condition. Therefore, some adaptive control schemes are needed for the MNS. Furthermore, the pneumatic actuator has excellent compliance and flexibility, and is suitable for the human-mechanical system. However, it also has nonlinearity, hence high precision control is quite difficult. Therefore, in this paper, the CMAC-PID control scheme is installed on the H8 microcomputer, and the adaptability of the control scheme is evaluated on some control experiments.

Flicker is one of the artifacts of MPEG-2 coded video caused by the display's gamma characteristics and depends on the picture texture. In order to improve the picture quality, we have developed a technique of reducing the flicker by display gamma simulations in the decoded pictures. The effectiveness was confirmed by experiments. The process is less complex, and the implementation of this technique in digital TV receivers is expected to improve the coded picture quality.

With advance of an aging society, the persons who are physically handicapped have their respective needs about mobility assist with their living conditions. Moreover, operating an electric wheelchair indoors in confined spaces requires considerable skill. This paper presents an obstacle avoiding support system for an electric wheelchair, using reinforcement learning. The obstacle avoidance is semi-automatically supported by the minimum vector field histogram (MVFH) method. The MVFH modifies the user manipulation and assists the obstacle avoidance. In the proposed scheme, the modification rate is adjusted by the reinforcement learning according to the environment and the user condition. The newly proposed scheme is numerically evaluated on a simulation example.

We developed a posture control system for an electric wheelchair equipped with active suspension, using CMAC, which is a neural network type of control. Testing of the posture control of the wheelchair, which considers a learning results as an output to the target value inputted on-line using the learning function of CMAC (cerebellar model arithmetic controller), was performed. In this paper, testing of the learning control, which holds the posture horizontally both by the simulation and in the experiment by the system, is performed, and the validity of CMAC is verified compared with conventional method.

We developed a posture control system for an electric wheelchair equipped with active suspension, using CMAC which is a neural network type of control. Testing of the posture control of the wheelchair, which considers a learning result as an output to the target value inputted on-line using the learning function of CMAC (cerebellar model arithmetic controller), was performed. In this paper, testing of the learning control which holds the posture horizontally both by the simulation and in the experiment by the system, is performed, and the validity of CMAC is verified.

The existing method for establishing travel routes provides modeled environmental information, but it is difficult to create an environment model for the environments where electric wheelchairs travel because the environment changes constantly due to the existence of moving objects including pedestrians. In this study, we propose an automatic travelling system for an electric wheelchair using reinforcement learning systems and CMACs. We select the best travel route by utilizing these reinforcement learning systems. When a CMAC learns the value function of Q-learning, an improved learning speed is achieved by utilizing the generalizing action. CMACs enable one to reduce the time needed to select the best travel route. Using simulation, a path planning experiment was performed.

We propose a new feature dimension reduction method for multimedia search. The main technique in the method is dynamic segmentation that partitions sequential feature trajectories dynamically. While dynamic segmentation reduces the average dimensionality and accelerates the search, it requires huge amount of calculation. Thus, our method quickly executes suboptimal partitioning of the trajectories by using the discreteness of dimension changes. This guarantees the optimal amount of calculation to derive the suboptimal partitioning under the condition that the dimension monotonously increases as the segment length increases. The experiment shows that our method is over 10 times faster than a straightforward dynamic segmentation method.

This paper describes a 250 kW hybrid system composed of two systems which uses clean energy sources, a wind power system and a photovoltaic (PV) system. The hybrid system has been installed in the radio relay station by NTT in Kume-jima (a small island in the south-west of Japan). The hybrid system consists of a 230 kW wind turbine, a 50 kW resistor, a 230 kW bi-directional converter with storage batteries, 20 kW PV modules, and a 20 kW interactive inverter. The power generated in this system is supplied to telecommunications equipment at the radio relay station and to the power grid by reversing power flow. One of the advantages of this system is that it suppresses rapid change in the output power of the wind power system. This rapid change causes the grid voltage and the grid frequency to fluctuate because the grid capacity in Kume-jima is low.

Previously, we proposed a histogram-based quick signal search method called Time-Series Active Search (TAS). TAS is a method of searching through long audio or video recordings for a specified segment, based on signal similarity. TAS is fast; it can search through a 24-hour recording in 1 second after a query-independent preprocessing. However, an even faster method is required when we consider a huge amount of audio archives, for example a month's worth of recordings. Thus, we propose a preprocessing method that significantly accelerates TAS. The core part of this method comprises a global histogram clustering of long signals and a pruning scheme using those clusters. Tests using broadcast recording indicate that the proposed algorithm achieves a search speed approximately 3 to 30 times faster than TAS. In these tests, the search results are exactly the same as with TAS.

We propose a quick and accurate search method for detecting a query signal from long video recordings The method is based on the time-series active search, which is a quick searching method for audio and video signals that we previously proposed. Time-series active search is based on a histogram matching scheme and an efficient pruning mechanism, and therefore, it was very quick. We found, however, that the accuracy sometimes deteriorates when it is applied to searches through long video archives that are composed of many similar video images or those containing feature distortions caused by video dubbing or low-bit-rate compression. The problem arises from (1) insufficient capability of representing features and (2) feature distortions. Thus, the method proposed here uses LBG-based VQ to improve the capacity to represent features and probabilistic dither-voting to improve robustness with respect to feature distortions. The experiments prove the effects of the proposed method.

We propose a quick algorithm for multimedia signal search. The algorithm comprises two techniques: feature compression based on piecewise linear maps and distance bounding to efficiently limit the search space. When compared with existing multimedia search techniques, they greatly reduce the computational cost required in searching. Although feature compression is employed in our method, our bounding technique mathematically guarantees the same recall rate as the search based on the original features; no segment to be detected is missed. Experiments indicate that the proposed algorithm is approximately 10 times faster than and as accurate as an existing fast method maitaining the same search accuracy.