Satellite Image Database Search Engine which Allows Fuzzy Expression of Geophysical Parameters of Queries

Satellite image database search engine which allows fuzzy expression of geophysical parameters of queries is proposed. A search engine based on knowledge based system which allows fuzzy expression of queries is proposed. A prototype system is created and tested. Whereas conventional search systems had to know in advance the functions of the search system, information about search keys, etc., the search engine proposed here guided the search conditions in a conversational form, by allowing ambiguous expressions (six adverb language hedges) at that time, the user is released from such annoyance. To make this possible, a membership function for each attribute information is defined, and a search condition refinement by fuzzy logic is introduced. The results show that the system accepts a fuzzy expression of query as well as a comprehensive dialogues between users and the system. Keywords—Search engine; fuzzy expression; knowledge base system; membership function


I. INTRODUCTION
The relational database (relational database) is frequently used for searching the earth observation satellite image database from the viewpoint of its efficiency and system flexibility. A typical example of the database search is a search engine using SQL (Structured Query Language) or QBE (Query by Example) [1], [2]. When these search engines are used, constructing search conditions for the earth observation satellite image database search, it is necessary to input a specific search key in a predetermined format. That is, the search can be performed only after the user sets the search condition according to the determined format. Therefore, it is necessary for the user to know in advance the functions of the search system, information on the search key.
The search system for the Earth Observation Satellite Image Database: EOSID proposed in this paper is based on SQL (since most existing EOSID systems have been developed on the premise of SQL), and allows search requests in natural language. The proposed search engine system translates the search request into a request in the SQL language (agent function [3]) and searches for the desired data. At the search engine is used as EOSID, the system guides the user in a conversational manner (guide system) and also to allow search requests with ambiguous expressions of natural language [4]. Therefore, a membership function for ambiguous expressions is introduced based on the opinions of users in the marine research field (for instance) and enabled fuzzy search. In order to confirm the effectiveness of the proposed method, a virtual database was constructed in the marine research field. Some experiments are performed with the prototype of the proposed search engine system. As a result, it is found that the search performance is quite good from the viewpoint of easiness of database retrievals.
In the following section, related research works are described. Then, the proposed prediction method and system is described followed by experimental set-up together with experimental results. After that, concluding remarks and some discussions are described.

II. RELATED RESEARCH WORKS
Vague search of earth observation image database based on Fuzzy theory using physical quantities and spatial features is proposed [5] together with earth observation satellite image database system allowing ambiguous search requests [6]. On the other hand, user friendly and efficient catalog information management for earth observation data, is proposed and well reported [7].
Remote sensing satellite image database system allowing image portion retrievals utilizing principal component which consists spectral and spatial features extracted from imagery data is proposed [8]. Meanwhile, data collection and active database for tsunami warning system is proposed [9]. On the other hand, remote sensing satellite image database system allowing image portion retrievals utilizing principal component which consists spectral and spatial features extracted from imagery data is proposed [10]. Also, numerical representation of web sites of remote sensing satellite data providers and its application to knowledge based information retrievals with natural language is proposed [11].
Image retrieval based on color, shape and texture for ornamental leaf with medicinal functionality, meanwhile, is proposed [12]. Also, comparison contour extraction based on layered structure and Fourier descriptor on image retrieval is proposed and evaluated its effectiveness [13]. Also, image retrieval method utilizing texture information derived from Discrete Wavelet Transformation: DWT together with color information is proposed [14].
Metadata definition and retrieval of earth observation satellite data is proposed [15]. Meanwhile, Geographic Information System: GIS based on neural network for www.ijacsa.thesai.org appropriate parameter estimation of geophysical retrieval equations with satellite remote sensing data is proposed [16]. Image retrieval method based on hue information and wavelet description based shape information as well as texture information of the objects extracted with dyadic wavelet transformation is proposed [17]. Wavelet based image retrievals is attempted [18]. Also, image retrieval method based on back-projection is proposed [19]. Fig. 1 shows the configuration of the proposed earth observation satellite image database system. The system allows search requests in natural language. Since the research on natural language understanding has not been completed, a text search method has been introduced here. That is, the text matching between the search request from the user and the search keyword prepared by the system is determined by Z39.504, [20].

A. System Configuration
A natural language whose expression is ambiguous has a broad meaning and is ambiguous. For example, a search may be requested in a vague image of the user, such as "slightly warm sea area" or "image with extremely small amount of cloud". Existing database search engines cannot handle such cases. In order to deal with this, the following fuzzy search is proposed.

B. Fuzzy Retrieval
A fuzzy search for users in the marine field is shown as an example of an ambiguous search. The search target (object [21] of the constructed virtual database in the marine field is an image ID (Identification No.), and the attribute information (Attributes) serving as a search key is as follows.

C. Creating Membership Functions
A fuzzy set (membership function [22], [23], [24]) was created to express the user's ambiguity regarding the abovementioned sea surface temperature, chlorophyll-a, and cloudiness. At that time, a questionnaire survey was conducted for researchers in the marine field of the Earth Environment Observation Committee of the Earth Science and Technology Agency, and based on the results (shown in Table I).
Equations (1) to (6) show the membership functions corresponding to low, high, high and low chlorophyll-a concentrations and high and low cloudiness, respectively.
where, the units of the physical quantity (x) are sea surface temperature (° C.), chlorophyll-a (mg / m 3 ), and cloudiness (%), respectively. Next, a linguistic expression (language hedge) that modifies the fuzzy set is defined. As a result, expressions such as "slightly warm" and "very" warm sea surface temperature can be accepted as search requests. The following shows the language hedge used in this prototype system.
 very y  more or less y  slightly y  sort of y  pretty y  rather y where y is a fuzzy set. logical operations using these combinations are also possible, which allows for modification of complex fuzzy sets, and allows search requests using combinations such as "slightly warm, chlorophyll-a sea areas ''.  Fig. 2 shows the search procedure. First,

1) A search request containing an ambiguous expression occurs,
2) Guide the search conditions in a conversational manner, 3) Select the membership function using the knowledge corresponding to the search condition, 4) Using the membership function in the database and then calculate the degree of attribution of data attributes, 5) Display search results.
where, the knowledge corresponding to the search condition is the semantic network shown in Fig. 3. The determination of the degree of belonging is performed as follows. In other words, if the attribute information of the data in the database is between a certain threshold value and the maximum value of the membership function, it is said to belong, and it is determined that the others do not belong. The determination of the threshold was determined as the grade in the center of gravity (CG) of the fuzzy set shown in equation (7).
where, A is an attribute item, and x is an attribute value (physical quantity).

E. AND, OR Logical Operation of Search Condition
Logical operations (AND, OR) of a plurality of search conditions are also possible. At this time, the degree of membership is given by equation (8).
www.ijacsa.thesai.org In other words, when searching for data with a slightly higher sea surface temperature and a very small amount of cloud, use Equation (8) that sets the two search conditions to AND. If any of the conditions is satisfied but data is desired, equation (9) may be used.

IV. EXPERIMENTS
Considering that sea surface temperature, chlorophyll-a, and cloudiness are uniformly distributed in the range of 0 to 40 (° C), 0 to 5 (mg / m 3 ), and 0 to 100 (%), respectively, 1000 pseudo Generated attribute information of earth observation data. Since the purpose here is to evaluate the performance of the ambiguous search, it is assumed that the search has already been completed for attributes other than these three items. Fig. 4 shows the search flow and search results when the following search request is entered.
In Fig. 4, the sentences with bold and underline characters indicates system response. Firstly, the system askes query, then user account and password are asked. After that, SST, chlorophyll-a concentration and cloud coverage are asked by the system. Then the search result with the table which is shown in Fig. 4 is returned from the system. If the user satisfies the search result, then the system asks whether or not the user would like to save the result. After that, the system asks user's willingness on other search request. If the user has no other search request, then the system is closed.
As a result, candidates meeting these search conditions (in this case, seven candidates) were presented, and it was confirmed that the user could further narrow down from these candidates. Experimental result shows the prediction accuracy of retrieve items are 100% of success rate. In terms of functionality of search engine of the current Earth Observation Satellite Image Database Systems, there is no geophysical parameter search based on knowledgebase system with fuzzy search capability. Therefore, the proposed search engine system has the merit on the geophysical research function. :n Now you may exit this system V. CONCLUSION A search engine based on knowledge based system which allows Fuzzy expression of queries is proposed. A prototype system is created and tested. Whereas conventional search systems had to know in advance the functions of the search system, information about search keys, etc., the search engine proposed here guided the search conditions in a conversational form, by allowing ambiguous expressions (six adverb language hedges) at that time, the user is released from such annoyance.
To make this possible, a membership function for each attribute information is defined, and a search condition refinement by fuzzy logic is introduced. The results show that the system accepts a Fuzzy expression of query as well as a www.ijacsa.thesai.org comprehensive dialogues between users and the system. As a result, candidates meeting these search conditions (in this case, seven candidates) were presented, and it was confirmed that the user could further narrow down from these candidates.

VI. FUTURE RESEARCH WORKS
The proposed system is adopted in the real earth observation satellite database system, and it is a future subject to realize a more usable database system.