Shinpei HAYASHI Keisuke ASANO Motoshi SAEKI
Goal refinement is a crucial step in goal-oriented requirements analysis to create a goal model of high quality. Poor goal refinement leads to missing requirements and eliciting incorrect requirements as well as less comprehensiveness of produced goal models. This paper proposes a technique to automate detecting bad smells of goal refinement, symptoms of poor goal refinement. At first, to clarify bad smells, we asked subjects to discover poor goal refinement concretely. Based on the classification of the specified poor refinement, we defined four types of bad smells of goal refinement: Low Semantic Relation, Many Siblings, Few Siblings, and Coarse Grained Leaf, and developed two types of measures to detect them: measures on the graph structure of a goal model and semantic similarity of goal descriptions. We have implemented a supporting tool to detect bad smells and assessed its usefulness by an experiment.
Liping ZHANG Zongqing LU Qingmin LIAO
This paper proposes a new and effective convolutional neural network model termed OFR-Net for optical flow refinement. The OFR-Net exploits the spatial correlation between images and optical flow fields. It adopts a pyramidal codec structure with residual connections, dense connections and skip connections within and between the encoder and decoder, to comprehensively fuse features of different scales, locally and globally. We also introduce a warp loss to restrict large displacement refinement errors. A series of experiments on the FlyingChairs and MPI Sintel datasets show that the OFR-Net can effectively refine the optical flow predicted by various methods.
Jeong-Uk BANG Mu-Yeol CHOI Sang-Hun KIM Oh-Wook KWON
As deep learning-based speech recognition systems are spotlighted, the need for large-scale speech databases for acoustic model training is increasing. Broadcast data can be easily used for database construction, since it contains transcripts for the hearing impaired. However, the subtitle timestamps have not been used to extract speech data because they are often inaccurate due to the inherent characteristics of closed captioning. Thus, we propose to build a large-scale speech database from multi-genre broadcast data with inaccurate subtitle timestamps. The proposed method first extracts the most likely speech intervals by removing subtitle texts with low subtitle quality index, concatenating adjacent subtitle texts into a merged subtitle text, and adding a margin to the timestamp of the merged subtitle text. Next, a speech recognizer is used to extract a hypothesis text of a speech segment corresponding to the merged subtitle text, and then the hypothesis text obtained from the decoder is recursively aligned with the merged subtitle text. Finally, the speech database is constructed by selecting the sub-parts of the merged subtitle text that match the hypothesis text. Our method successfully refines a large amount of broadcast data with inaccurate subtitle timestamps, taking about half of the time compared with the previous methods. Consequently, our method is useful for broadcast data processing, where bulk speech data can be collected every hour.
Shinnosuke SARUWATARI Fuyuki ISHIKAWA Tsutomu KOBAYASHI Shinichi HONIDEN
Refinement-based formal specification is a promising approach to the increasing complexity of software systems, as demonstrated in the formal method Event-B. It allows stepwise modeling and verifying of complex systems with multiple steps at different abstraction levels. However, making changes is more difficult, as caution is necessary to avoid breaking the consistency between the steps. Judging whether a change is valid or not is a non-trivial task, as the logical dependency relationships between the modeling elements (predicates) are implicit and complex. In this paper, we propose a method for analyzing the impact of the changes of Event-B. By attaching labels to modeling elements (predicates), the method helps engineers understand how a model is structured and what needs to be modified to accomplish a change.
Jun WANG Lei HU Ning LI Chang TIAN Zhaofeng ZHANG Mingyong ZENG Zhangkai LUO Huaping GUAN
This paper presents a novel model in the field of image co-saliency detection. Previous works simply design low level handcrafted features or extract deep features based on image patches for co-saliency calculation, which neglect the entire object perception properties. Besides, they also neglect the problem of visual similar region's mismatching when designing co-saliency calculation model. To solve these problems, we propose a novel strategy by considering both local prediction and global refinement (LPGR). In the local prediction stage, we train a deep convolutional saliency detection network in an end-to-end manner which only use the fully convolutional layers for saliency map prediction to capture the entire object perception properties and reduce feature redundancy. In the global refinement stage, we construct a unified co-saliency refinement model by integrating global appearance similarity into a co-saliency diffusion function, realizing the propagation and optimization of local saliency values in the context of entire image group. To overcome the adverse effects of visual similar regions' mismatching, we innovatively incorporates the inter-images saliency spread constraint (ISC) term into our co-saliency calculation function. Experimental results on public datasets demonstrate consistent performance gains of the proposed model over the state-of-the-art methods.
Tetsunao MATSUTA Tomohiko UYEMATSU
In the successive refinement problem, a fixed-length sequence emitted from an information source is encoded into two codewords by two encoders in order to give two reconstructions of the sequence. One of two reconstructions is obtained by one of two codewords, and the other reconstruction is obtained by all two codewords. For this coding problem, we give non-asymptotic inner and outer bounds on pairs of numbers of codewords of two encoders such that each probability that a distortion exceeds a given distortion level is less than a given probability level. We also give a general formula for the rate-distortion region for general sources, where the rate-distortion region is the set of rate pairs of two encoders such that each maximum value of possible distortions is less than a given distortion level.
Tetsunao MATSUTA Tomohiko UYEMATSU
This paper deals with a broadcast network with a server and many users. The server has files of content such as music and videos, and each user requests one of these files, where each file consists of some separated layers like a file encoded by a scalable video coding. On the other hand, each user has a local memory, and a part of information of the files is cached (i.e., stored) in these memories in advance of users' requests. By using the cached information as side information, the server encodes files based on users' requests. Then, it sends a codeword through an error-free shared link for which all users can receive a common codeword from the server without error. We assume that the server transmits some layers up to a certain level of requested files at each different transmission rate (i.e., the codeword length per file size) corresponding to each level. In this paper, we focus on the region of tuples of these rates such that layers up to any level of requested files are recovered at users with an arbitrarily small error probability. Then, we give inner and outer bounds on this region.
Kojiro TAKEYAMA Satoshi MAKIDO Yoshiko KOJIMA
In recent years, various Portable Navigation Devices (PND) such as smart-phones are becoming popular as a vehicle navigation device. To compare with a conventional built-in navigation system, PND has advantages that it is low cost and easily mounted to the vehicle. On the other hand, PND has also disadvantage that in the most case it cannot obtain the reliable vehicle speed information such as wheel pulse information and that induces degradation of vehicle trajectory estimation (dead-reckoning). The vehicle trajectory estimation is the positioning method using inertial sensors, and generally used when GPS is not available. So in urban area where GPS signals are blocked or reflected by tall buildings, the degradation of vehicle trajectory estimation may cause the severe increase of position error. Accordingly, in this study two approaches are proposed to improve vehicle trajectory estimation with PND. The first one is the accurate speed estimation using time-series tightly coupled integration of accelerometer, gyro, and Doppler shift of GPS. And the second one is the correction of trajectory error using backward refinement that can work even in real-time processing. The experimental result in Shinjuku which is dense urban city shows that the error of vehicle trajectory estimation was reduced to 1/4 compared with the previous method.
Yongxin ZHAO Yanhong HUANG Qin LI Huibiao ZHU Jifeng HE Jianwen LI Xi WU
Survivability is an essential requirement of the networked information systems analogous to the dependability. The definition of survivability proposed by Knight in [16] provides a rigorous way to define the concept. However, the Knight's specification does not provide a behavior model of the system as well as a verification framework for determining the survivability of a system satisfying a given specification. This paper proposes a complete formal framework for specifying and verifying the concept of system survivability on the basis of Knight's research. A computable probabilistic model is proposed to specify the functions and services of a networked information system. A quantified survivability specification is proposed to indicate the requirement of the survivability. A probabilistic refinement relation is defined to determine the survivability of the system. The framework is then demonstrated with three case studies: the restaurant system (RES), the Warship Command and Control system (LWC) and the Command-and-Control (C2) system.
Moving objects or more generally foreground objects are the simplest objects in the field of computer vision after the pixel. Indeed, a moving object can be defined by 4 integers only, either two pairs of coordinates or a pair of coordinates and the size. In fixed camera scenes, moving objects (or blobs) can be extracted quite easily but the methods to produce them are not able to tell if a blob corresponds to remaining background noise, a single target or if there is an occlusion between many target which are too close together thus creating a single blob resulting from the fusion of all targets. In this paper we propose an novel method to refine moving object detection results in order to get as many blobs as targets on the scene by using a tracking system for additional information. Knowing if a blob is at proximity of a tracker allows us to remove noise blobs, keep the rest and handle occlusions when there are more than one tracker on a blob. The results show that the refinement is an efficient tool to sort good blobs from noise blobs and accurate enough to perform a tracking based on moving objects. The tracking process is a resolution free system able to reach speed such as 20 000fps even for UHDTV sequences. The refinement process itself is in real time, running at more than 2000fps in difficult situations. Different tests are presented to show the efficiency of the noise removal and the reality of the independence of the refinement tracking system from the resolution of the videos.
Bei HE Guijin WANG Chenbo SHI Xuanwu YIN Bo LIU Xinggang LIN
Based on sample-pair refinement and local optimization, this paper proposes a high-accuracy and quick matting algorithm. First, in order to gather foreground/background samples effectively, we shoot rays in hybrid (gradient and uniform) directions. This strategy utilizes the prior knowledge to adjust the directions for effective searching. Second, we refine sample-pairs of pixels by taking into account neighbors'. Both high confidence sample-pairs and usable foreground/background components are utilized and thus more accurate and smoother matting results are achieved. Third, to reduce the computational cost of sample-pair selection in coarse matting, this paper proposes an adaptive sample clustering approach. Most redundant samples are eliminated adaptively, where the computational cost decreases significantly. Finally, we convert fine matting into a de-noising problem, which is optimized by minimizing the observation and state errors iteratively and locally. This leads to less space and time complexity compared with global optimization. Experiments demonstrate that we outperform other state-of-the-art methods in local matting both on accuracy and efficiency.
Hisashi MIYAZAKI Tomoyuki YOKOGAWA Sousuke AMASAKI Kazuma ASADA Yoichiro SATO
During a software development phase where a product is progressively elaborated, it is difficult to guarantee that the refined product retains its original behaviors. In this paper, we propose a method to detect refinement errors in UML sequence diagrams using LTSA (Labeled Transition System Analyzer). The method integrates multiple sequence diagrams using hMSC (high-level Message Sequence Charts) into a sequence diagram. Then, the method translates the diagram into FSP representation, which is the input language of LTSA. The method also supports some combined fragment operators in the UML 2.0 specification. We applied the method to some examples of refined sequence diagrams and checked the correctness of refinement. As a result, we confirmed the method can detect refinement errors in practical time.
Chao LIAO Guijin WANG Bei HE Chenbo SHI Yongling SHEN Xinggang LIN
The time efficiency of aerial video stitching is still an open problem due to the huge amount of input frames, which usually results in prohibitive complexities in both image registration and blending. In this paper, we propose an efficient framework aiming to stitch aerial videos in real time. Reasonable distortions are allowed as a tradeoff for acceleration. Instead of searching for globally optimized solutions, we directly refine frame positions with sensor data to compensate for the accumulative error in alignment. A priority scan method is proposed to select pixels within overlapping area into the final panorama for blending, which avoids complicated operations like weighting or averaging on pixels. Experiments show that our method can generate satisfying results at very competitive speed.
Chan-Hee HAN Si-Woong LEE Hamid GHOLAMHOSSEINI Yun-Ho KO
In this paper, side information refinement methods for Wyner-Ziv video codec are presented. In the proposed method, each block of a Wyner-Ziv frame is separated into a predefined number of groups, and these groups are interleaved to be coded. The side information for the first group is generated by the motion compensated temporal interpolation using adjacent key frames only. Then, the side information for remaining groups is gradually refined using the knowledge of the already decoded signal of the current Wyner-Ziv frame. Based on this basic concept, two progressive side information refinement methods are proposed. One is the band-wise side information refinement (BW-SIR) method which is based on transform domain interleaving, while the other is the field-wise side information refinement (FW-SIR) method which is based on pixel domain interleaving. Simulation results show that the proposed methods improve the quality of the side information and rate-distortion performance compared to the conventional side information refinement methods.
Lei WANG Jun WANG Satoshi GOTO Takeshi IKENAGA
With the ubiquitous application of Internet and wireless networks, H.264 video communication becomes more and more common. However, due to the high-efficiently predictive coding and the variable length entropy coding, it is more sensitive to transmission errors. The current error concealment (EC) scheme, which utilizes the spatial and temporal correlations to conceal the corrupted region, produces unsatisfied boundary artifacts. In this paper, first we propose variable block size error concealment (VBSEC) scheme inspired by variable block size motion estimation (VBSME) in H.264. This scheme provides four EC modes and four sub-block partitions. The whole corrupted macro-block (MB) will be divided into variable block size adaptively according to the actual motion. More precise motion vectors (MV) will be predicted for each sub-block. Then MV refinement (MVR) scheme is proposed to refine the MV of the heterogeneous sub-block by utilizing three step search (TSS) algorithm adaptively. Both VBSEC and MVR are based on our directional spatio-temporal boundary matching algorithm (DSTBMA). By utilizing these schemes, we can reconstruct the corrupted MB in the inter frame more accurately. The experimental results show that our proposed scheme can obtain better objective and subjective EC quality, respectively compared with the boundary matching algorithm (BMA) adopted in the JM11.0 reference software, spatio-temporal boundary matching algorithm (STBMA) and other comparable EC methods.
Erin-Ee-Lin LAU Wan-Young CHUNG
A novel RSSI (Received Signal Strength Indication) refinement algorithm is proposed to enhance the resolution for indoor and outdoor real-time location tracking system. The proposed refinement algorithm is implemented in two separate phases. During the first phase, called the pre-processing step, RSSI values at different static locations are collected and processed to build a calibrated model for each reference node. Different measurement campaigns pertinent to each parameter in the model are implemented to analyze the sensitivity of RSSI. The propagation models constructed for each reference nodes are needed by the second phase. During the next phase, called the runtime process, real-time tracking is performed. Smoothing algorithm is proposed to minimize the dynamic fluctuation of radio signal received from each reference node when the mobile target is moving. Filtered RSSI values are converted to distances using formula calibrated in the first phase. Finally, an iterative trilateration algorithm is used for position estimation. Experiments relevant to the optimization algorithm are carried out in both indoor and outdoor environments and the results validated the feasibility of proposed algorithm in reducing the dynamic fluctuation for more accurate position estimation.
Yutao DONG Xiangzhong FANG Jing YANG
This letter proposes a new algorithm of refining the quantization parameter in H.264 real-time encoding. In the H.264 encoding, the quantization parameter computed according to the quadratic rate model is not accurate in meeting the target bit rate. In order to make the actual encoded bit rate closer to the target bit rate, ρ-domain rate model is introduced in our proposed quantization parameter refinement algorithm. Simulation results show that the proposed algorithm achieves obvious gain in PSNR and has stabler encoded bit rate compared to Jiang's algorithm.
A motion refinement algorithm is proposed to enhance motion compensated noise reduction (MCNR) efficiency. Instead of the vector with minimum distortion, the vector with minimum distance from motion vectors of neighboring blocks is selected as the best motion vector among vectors which have distortion values within the range set by noise level. This motion refinement finds more accurate motion vectors in the noisy sequences. The MCNR with the proposed algorithm maintains the details of an image sequence very well without blurring and joggling. And it achieves 10% bit-usage reduction or 0.5 dB objective quality enhancement in subsequent video coding.
Motion-compensated frame interpolation (MCFI) is widely used to smoothly display low frame rate video sequences by synthesizing and inserting new frames between existing frames. The temporal shift interpolation technique (TSIT) is popular for frame interpolation of video sequences that are encoded by a block-based video coding standard such as MPEG-4 or H.264/AVC. TSIT assumes the existence of a motion vector (MV) and may not result in high-quality interpolation for intra-mode blocks that do not have MVs. This paper proposes a new frame interpolation algorithm mainly designed for intra-mode blocks. In order to improve the accuracy of pixel interpolation, the new algorithm proposes sub-pixel interpolation and the reuse of MVs for their refinement. In addition, the new algorithm employs two different interpolation modes for inter-mode blocks and intra-mode blocks, respectively. The use of the two modes reduces ghost artifacts but potentially increases blocking effects between the blocks interpolated by different modes. To reduce blocking effects, the proposed algorithm searches the boundary of an object and interpolates all blocks in the object in the same mode. Simulation results show that the proposed algorithm improves PSNR by an average of 0.71 dB compared with the TSIT with MV refinement and also significantly improves the subjective quality of pictures by reducing ghost artifacts.
Tae Meon BAE Truong Cong THANG Yong Man RO
In this letter, we propose an enhanced method for inter-layer motion prediction in scalable video coding (SVC). For inter-layer motion prediction, the use of refined motion data in the Fine Granular Scalability (FGS) layer is proposed instead of the conventional use of motion data in the base quality layer to reduce the inter-layer redundancy efficiently. Experimental results show that the proposed method enhances coding efficiency without increasing the computational complexity of the decoder.