So far, Automatic Signature Verification (ASV) approaches using a threshold-based decision have depended on one feature set for distance measure and a threshold on this distance measure for verification. The best performance that can be reached in this case is the one obtained by using the best feature set (bfs). In this paper, we introduce a new decision making approach for ASV that uses Multi-Sets of Features (MSF). The MSF provides higher performance than that obtainable by using the bfs, with better forgery detection. The improvement is seen to be significant because it recovers some lost effectiveness and can add it to that of the bfs. This gain in effectiveness is highly desirable when we deal with signatures of high value documents.

Networked video delivering systems have been developing. When videos are delivered from servers to client sites. Moreover, link overflow does not occur, since capability of links are restricted. The systems allocate resources for meeting these requirements. For effective delivery, it is necessary that a policy of scheduling be established. A proposed policy is that the video is transmitted with the minimum bandwidth. The minimum bandwidth is corresponding to a consuming ratio of the video. The scheduling method consists of three steps. They are to find paths for delivering the video clips, to calculate required time to deliver them without link overflow, and to set their start times from servers without skew time. Moreover, simulation of this method is achieved by making schedules to deliver video. This policy is simple and flexible to deliver videos.

A news browsing system has been developed based on caricatures in newspapers, called Picture Roll. If news is significant, a caricature is depicted in the newspaper for explaining it. Moreover, some news stories are written, and they are related to the caricature. To browse the significant news, it is suitable to show caricatures and related news stories. Picture Roll has been developed for providing such browsing functions. The system provides two browsing facilities which are based on clusters and timeline. In the former, keywords are assigned to each caricature, The caricatures are classified into the space that is constructed by the latent semantic indexing. In the latter, caricatures are related to, each other by links representing the time-line. News items are shown in order of time. Users are able to browse the news stories by navigating caricatures.

This paper presents mechanisms for annotating and indexing video by keywords in the Video Object Management system, called VOM. In order to annotate video, an entire video is divided into its video clips, and keywords are assigned to them. These clips are represented in objects, and the objects are organized into a hierarchical data structure in sense of their playback order. Moreover, such objects include a collection of keywords for specifying each object. Then, one object is defined to concatenate several short clips. Keywords assigned to lower positioned objects are abstracted in proportion as this concatenation. The keywords of higher positioned objects are chosen from a collection of the keywords that are assigned into lower ones, or are associated from a set of existing keywords. VOM provides a mechanism to assist in keyword abstraction based on classification of keywords and objects. A collection of them is classified by latent semantic indexing. In addition, the classes consisting of the objects and their keywords are reflected the playback order of the objects.

We present a motion analysis system of soccer games. The purpose of this system is to evaluate the teamwork quantitatively based on the movement of all the players in a game. Space management and cooperative movement by the players are two major factors for teamwork evaluation. To quantify them from motion images, we propose two new features; "minimum moving time pattern" and "dominant region". From experiments using actual game scenes, it is suggested that the proposed system can be a new tool for supporting to evaluate the teamwork.

We address the knowledge representation for understanding table-form documents. In particular, we focus on the knowledge for establishing the correctness of recognized results in addition to the traditional syntactic/constructive-oriented understanding framework. In order to attain to this subject, we introduce the semantic information to specify domains and relationships among individual items definitely, which is complementary to the syntactic/constructive information available in the traditional framework.

In order to integrate a knowledge-based system (KBS) and relational database system (RDBS) more naturally and flexibly, it is necessary to work with these two systems cooperatively. The IKD system functions as an interface between the KBS and RDBS by representing their conceptual structures in frames. Two types of frames are distinguished from one another: one defines the data structure of an integrating database (DB) and the other represents conceptual objects dealing with a KBS. Several types of operations on the DB are processed based on message passing among frames.>

The authors propose a fundamental framework for the editing facility applicable to uniform manipulation for various kinds of data. The basic principle is to separate the object scheme and the data instance from the editing data extractively. This separation method makes it possible to specify various kinds of data under explicitly structured forms and then manipulate them with the uniform access paths. The editing model has an object-oriented feature based on the relational data model. The design concept of the editing mechanism is described in detail. Integration and functionality issues are addressed.>

A method for signature description based on the human way of recognizing objects is introduced. Like humans, this method first recognizes objects globally and then recognizes finer details. The description is hierarchical: the signature is at the top level, and each lower level gives more details about the higher one. Features are extracted in a relative way (e.g. the length is measured by its ratio to the width) to make the method insensitive to the natural variations that occur in genuine signatures, and are mapped into a conceptual description. This conceptual description is then used to classify the signature images.>

This paper introduces an effective method for signature separation from nonhomogeneous noisy background. It also introduces a solution to the problem of simulated signature verification in off-line systems. Extraction of shape and density features and the effectiveness of using each and both of them are discussed in the light of experimental results.

In this paper, we present a generalization of the distance transformation of a digitized picture in two different aspects. First, we define the generalized distance transformation of a binary picture (GDTB). A subclass of GDTB, called a local minimum filter family of GDTB (LMF-GDTB), characterized by a series of local minimum filters with varying neighborhoods, is discussed in detail. A skeleton is defined for LMF-GDTB, and it is proved that any binary picture can be reconstructed exactly from its skeleton with the distance value on it. Second, the gray weighted distance transformation (GWDT) is extended to a generalized GWDT (GGWDT) by introducing an arbitrary initial picture. After the fundamental equation of GGWDT and its solution are derived, it is proved that an arbitrary gray picture is generated by iterative application of GGWDT from a uniquely determined elementary picture and a sequence of initial value pictures.

The distance transformation of a line pattern (DTLP) is proposed and its properties are studied in detail. DTLP is the transformation of a binary line pattern by which the value at each 1-pixel in an input pattern is changed into the distance measured along a line pattern from that pixel to the furthest edge point (type I DTLP) or to the nearest edge point (type II DTLP). Both types of DTLP are performed by iterative local parallel operations. DTLP is effectively used in the processing of line patterns, such as the elimination or extraction of branches having prespecified features, and the description of the structure of a line pattern. The transformation can also be applied to the texture analysis of line patterns. Several parallel operations for feature extraction related to line patterns based upon the DTLP are given. Experimental results are shown concerning the application of the DTLP to the processing of blood vessel images in chest radiograms and to the texture analysis of photomicrographs of metal pieces. Line pattern analysis based on the DTLP is distinguished from other line pattern processing methods, such as graphical network analysis techniques, in that a two-dimensional line pattern can be processed directly by parallel local operations; and properties based on length as well as topology are effectively used.

Paz et al. [1] have revealed the necessary and sufficient conditions for a linear automaton to be strongly approximable. In this paper, we extend their theory to a linear space automaton [2], i.e., a linear system with more than one linear mapping, to obtain a necessary and sufficient condition that it is strongly stable.

Recently Kondratyev and Trofimov obtained a theorem for binaryANcodes with distance not less than five. In this correspondence we prove that this theorem holds for arbitrary nonbinary radix. Noting that the generalization of Peterson's theorem on single-error-correcting binaryANcodes to nonbinary cases does not seem to be easily forthcoming the result described here is surprising.