KR20000037798A - Expert system - Google Patents

Abstract

PURPOSE: An expert system is provided to reduce cost and effort for an image managing expert by constructing examples of development of automatic examination system through image management as knowledge base and by finding and showing same examples automatically when an user executes a specific examination work and by finding and showing the most similar example if same examples are not present in the knowledge base. CONSTITUTION: An expert system includes an example base(11) which subdivides applied examples into image management work units and stores them; an example managing module(13) which loads examples stored the example base(11) or stores supplied success examples or failure examples in the example base(11) when an user shows a problem; and a deduction engine(14) which searches the loaded examples from the example managing module(13) and verifies propriety and provides success examples or failure examples to the example managing module(13).

Description

Expert System

The present invention relates to an expert system, and in particular, by building a development case of an automatic inspection system through image processing as a knowledge base, when a user wants to perform a specific inspection task, the same case is automatically found and presented, and the same case is a knowledge base. If it does not exist, it is related to an expert system that deduces the application method by finding the most similar case.

In general, in order to build an automatic inspection system in the image processing field, an image processing expert is performed through various trial and error. Systems built through this process can only be used in that area and there are many problems with reuse.

The first is the absence of image processing experts. The number of image processing specialists is very small and the cost is high, so it is very expensive to build a new system. In addition, the development experience of image processing experts is also varied, it is one of the difficulties to have a significant difference in the individual experience and capabilities.

An object of the present invention is to build a case of development of an automatic inspection system through image processing as a knowledge base, when a user wants to perform a specific inspection task automatically finds the same case and presents the same case, if the same case does not exist in the knowledge base It is to provide an expert system that infers the application method by finding the most similar cases.

1 is a block diagram of an expert system according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a case stored in the case base of FIG. 1 in a tree structure. FIG.

FIG. 3 is a flow chart showing the inference process of the inference engine of FIG.

(Explanation of symbols for the main parts of the drawing)

11: Case Base 12: Knowledge Base

13: case management module 14: inference engine

15: Case Search Module 16: Case Suitability Testing Module

21: root nodes 22, 23, 24: task nodes

25-29: Unit of Work Node

Means for achieving the object of the present invention is a case base for subdividing the applied case into the image processing unit of work, and if a problem is presented by the user, loading the case stored in the case base, or provided success stories Or a case management module for storing failure cases in the case base, and an inference engine for providing the case management module with success cases or failure cases by retrieving and verifying the cases loaded by the case management module. do.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of an expert system according to an exemplary embodiment of the present invention.

As shown in FIG. 1, an expert system according to an embodiment of the present invention includes a case base 11 for subdividing and storing an applied case into image processing work units, and a knowledge base storing general knowledge related to image processing ( 12) and a case management module 13 for loading a case stored in the case base 11 or storing the provided success case or failure case in the case base 11 when a problem is presented by the user. , The reasoning engine 14 provides the case management module 13 with success cases or failure cases by searching and verifying the cases loaded by the case management module 13 having a case-based tree structure. .

The reasoning engine 14 is a case search module 15 for searching in the root node of the case-based tree whether there is a case that matches the presented problem, and a case where the search result in the case search module 15 matches exactly. There is no case conformance test module 16 for finding a matching feature in the task node or unit of work node of the case-based tree.

FIG. 2 is a tree structure illustrating a case stored in the case base of FIG. 1.

As shown in FIG. 2, the case-based tree includes a minimum image processing unit of work node (25, 26) and (27, 28), an intermediate task node (22, 23, 24) in which work units are grouped, and a case. It has a hierarchical structure composed of representative root nodes 21.

The minimum image processing work unit nodes 25, 26, 27, 28 and task nodes 22, 23, 24 have directionality and priority, respectively.

The operation of the expert system according to the embodiment of the present invention configured as described above will be described in detail with reference to FIGS. 2 and 3.

First, referring to FIG. 1, there is no proved example of whether image processing knowledge for constructing an automatic inspection system can be expressed as a general rule. On the other hand, since there are many cases, each case has a proven problem. The characteristic and its solution are stored in the case base 11 as an example. The basic knowledge related to image processing is stored and used in the additional knowledge base 12.

The case management module 13 stores the success cases or failure cases of the automatic inspection system using the image processing in the case base 11 or loads the cases stored in the case base 11.

The case search module 15 searches for a case that can solve the problem, and the case suitability test module 16 verifies suitability by finding a difference in a problem in the result searched in the case search module 15 to verify suitability. You will find a new year.

Each case stored in the case base 11 is subdivided and stored in the smallest image processing work unit. As shown in Fig. 2, one example is a hierarchical structure consisting of such minimum image processing working nodes 25, 26, 27, 28 and task unit nodes 22, 23, 24, in which the working nodes are grouped. It is represented as a tree. Several task nodes 22, 23, and 24 exist under the root node 21 representing a case, and directionality and priority exist between the nodes 22, 23, and 24. Below these task nodes 22, 23, 24 are the most basic image processing operations 24, 25, 27, 28, which also have directionality with each other.

Referring to Figure 3, the reasoning process of the inference engine is as follows.

First, in step S1, when a new problem is presented to the expert system by the user, in step S2, the expert system retrieves from the root node 21 of the case hierarchy tree shown in FIG. 2 whether there is a matching case. If there is a match in the search result in step S2, in steps S3 and S4, the expert system searches for the unit of work nodes 25, 26, 27, 28 of the case-based tree searched for by the user and presents the It gives a solution to the problem.

As a result of the search in step S2, if there is no exact match, in step S5, the system asks the user an additional question about the problem, and the task nodes 22, 23, 24 or unit of work in the case-based tree. Nodes 25, 26 and 27, 28 find matching features. This process is conformity verification.

As a result of conformity verification at step S5, if a matched feature is found, at step S8, the system presents a solution to the found node with the corresponding case to the user, and finds a new case hierarchy tree using the found partial solutions. Configure it and present it to the user.

In step S9, the user applies the problem through the proposed solution and informs the system that the case has been proved if the problem is properly solved, and in step S10 and S11, the system assigns an index to this case and returns to the case base ( 11).

As a result of conformity verification in step S5, if the presented case does not solve the problem, in step S6 and S7, the system constructs a new case configuration tree again and presents it to the user and at the same time verifies the failure case and the result Is stored in the failure case base 11 to be used for future performance improvement.

In this way, the Expert system constructs all the case building trees that it can build and presents the case to the user continuously until the problem is solved.

According to the expert system of the present invention, by replacing the image processing expert in the construction of a simple automatic inspection system, the cost and effort for the image processing expert can be reduced.

In addition, the present invention can be used to build a complex inspection system when the case is continuously stored and inferred in the expert system, and has the effect of generalizing and documenting the knowledge that image processing experts use in the construction of an automatic inspection system. have.

In addition, the present invention can be expected educational effect by explaining the problem solving step by step through the stored case.

Claims (5)

A case base for subdividing and applying the applied cases into image processing work units; A case management module for loading a case stored in the case base or storing the provided success case or failure case in the case base when a problem is presented by the user; And an inference engine configured to provide a success case or a failure case to the case management module by retrieving the case loaded by the case management module and verifying suitability. The method of claim 1, An example system stored in the case base has an case-based tree structure. The method of claim 2, The case-based tree has an hierarchical structure comprising a minimum image processing work unit node, a middle task node in which work units are grouped, and a root node representing a case. The method of claim 3, wherein And at least one of the minimum image processing work unit node and a task node has direction and priority. The method according to claim 1 or 3, The reasoning engine includes: a case searching module for searching at the root node of the case-based tree whether there are any cases that match the presented problem; Expert case, characterized in that consisting of a case suitability test module for finding a matching feature in a task node or a work unit node of the case-based tree, if there is no exact case of the search results in the case search module.