Mikihiko Nishiara

A Framework Toward Restoration of Writing Order from Single-Stroked Handwriting Image

Restoration of writing order from a single-stroked handwriting image can be seen as the problem of finding the smoothest path in its graph representation. In this paper, a 3-phase approach to restore a writing order is proposed within the framework of the edge continuity relation (ECR). In the initial, local phase, in order to obtain possible ECRs at an even-degree node, a neural network is used for the node of degree 4 and a theoretical approach is presented for the node of degree higher than 4 by introducing certain reasonable assumptions. In the second phase, we identify double-traced lines by employing maximum weighted matching. This makes it possible to transform the problem of obtaining possible ECRs at odd-degree node to that at even-degree node. In the final, global phase, we find all the candidates of single-stroked paths by depth first search and select the best one by evaluating SLALOM smoothness. Experiments on static images converted from online data in the Unipen database show that our method achieves a restoration rate of 96.0 percent

On the AEP of word-valued sources

We consider a new class of information sources called word-valued sources in order to investigate coding algorithms based upon string parsing. A word-valued source is defined as a pair of an independent and identically distributed (i.i.d.) source with a countable alphabet and a function that maps each symbol into a finite sequence over a finite alphabet. A word-valued source is a nonstationary process and has countable states. If the function of a word-valued source is prefix-free, the entropy rate is characterized with a simple expression and the AEP (asymptotic equipartition property) holds.

A novel approach to recover writing order from single stroke offline handwritten images

Problem of recovering the writing order from single-stroked handwritten image can be seen as finding the smoothest Euler path in its graph representation. In this paper, a novel approach is proposed to solve the recovery problem within the framework of the edge contiguous relation (ECR). Firstly, we make local analyses to obtain the possible ECRs at each of the nodes; secondly a global trace is executed to find all of the candidate Euler paths and the smoothest one is selected as a final result. Based on two simple assumptions, we prove a series of theorems to obtain possible ECRs at even node. Double-traced lines are identified by using the weighted matching of general graph. Experiments on the scanned images and offline images converted from the online data of Unipen database have shown that our method achieved 95.2% correct recovery rate.

Coding for Sources That Output Symbols According to Poisson Process*This work is supported in part by the grant-in-aid for scientific research, 16560326, 17360178 and 18760267.

We consider coding for sources that output the symbols according to Poisson process from the viewpoint of real-time transmission. In order to reduce the transmission delay we avoid using input buffers. However, the lack of buffer causes overflow error. The theoretical relation between the transmission rate and the error probability is clarified. It is shown that the optimal code that minimizes the probability of error differs from the code that minimizes the expected codeword length. We also investigate the case of block coding as one of the applications of buffers.

On relationship between loss probability and utilization factor on real-time basis

It is known that for sources that output symbols according to Poisson process the loss probability coincides with the utilization factor of the channel for any code on the realtime basis. It is also known for continuous-time Markov chains they take the same value for any code. This coincidence means that minimizing loss probability gives us the maximum spare time to transmit another information. That is, we can enjoy two merits by optimizing one parameter. In this paper, we specify the source class where we have such a benefit. Concretely, as a super class including Poisson processes and continuous-time Markov chains, we define a class of sources that traverse a finite state space with arbitrary random stay time. For this source class, we show that the necessary and sufficient condition for the coincidence of values between the loss probability and the utilization factor is that for each state the stay time obeys an exponential distribution. Assuming that the stay time obeys an exponential distribution, we derive the minimum loss probability. Moreover, we derive the minimum loss probability in the case that a single code is common in all states.

Multicast Communication for Video Broadcasting Service over IPv4 Network Using IP Option

In this paper, we propose a protocol for multicast communication called Multicast Datagram Transfer Protocol (MDTP) to provide multicast for video broadcasting service on the Internet. MDTP is a one-to-many multicast communication protocol, which is constructed based on IPv4 unicast protocol by utilizing IP Router Alert Option, and it uses unicast addressing and unicast routing protocol. A mechanism is presented to allow a router to remove identical video stream, to duplicate a video stream, and to forward each copy of the duplicated video stream to its destinations. Ordinary IP routers that do not support MDTP will treat the MDTP packets as normal unicast packets. Hence, gradual deployment is possible without tunneling technique. With a delegation mechanism, MDTP router is also able to handle request from clients, and serve the requested video stream. The simulation results show that the average bandwidth usage of MDTP is close to the average bandwidth usage of IP multicast. MDTP also has greater efficiency than XCAST, and its efficiency becomes significant for a large number of clients.

Study on Coding for Continuous-Time Markov Chains on Real-Time Basis

We consider on coding for continuous-time Markov chains from the viewpoint of real-time transmission. The theoretical relation between the transmission rate of the channel and the loss probability is described. It is shown that the channel use efficiency coincides with the loss probability.

On Multiple Smoothed Transmission of MPEG Video Streams

A rate splitting algorithm is presented for a multiple video transmission system to transfer the aggregation (or statical multiplexing) of multiple video streams to multiple clients so that each client can receive the requested video stream with the reliable fidelity. Computer simulations for transmission of a set of 128 MPEG compressed video streams show that the proposed algorithm alleviates the variability of the aggregate video transmission comparing with a scheme to smooth individually each of videos using the traditional online smoothing algorithm. Besides, the proposed is 2 time faster than the traditional one.

Multicast communication for scalable video application using IP option

In this paper, we are concerned with single source multicast communication models for scalable video application in IPv4 network using unicast protocol. We propose a new protocol called multicast datagram transfer protocol (MDTP) which utilizes the IP option field of the IPv4 header. MDTP enables routers responsible to replicate and forward MDTP packets, and to do control related tasks, such as adding or dropping video layers, and terminating flow to a client. Simulation results show that proposed protocol performs better on the average rate over another unicast based protocol called explicit multicast (Xcast). The proposed protocol also keeps only one information about the highest layer flowing through the router instead of addressing every layer when using the multicast protocol.

Achievability of maximum decoding rate on sequential coding

A sequential code such as an arithmetic code is a code that determines a prefix of the codeword from a prefix of the source sequence. The reciprocal of the time period in which the transmitter can send a single code symbol is called the transmission velocity. Since, at every moment, the restored length cannot be longer than the source length, the decoding rate is one at most. In this manuscript, we derive a transmission velocity at which the decoding rate is one almost surely with sequential codes.