JPH04180175A - Picture processor - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an image processing device for searching and retrieving necessary images.

BACKGROUND ART A conventional device of this kind is, for example, one shown in "Knowledge Engineering and Artificial Intelligence J, pp. 56-9, 65-72.

FIG. 2 shows a configuration diagram of this conventional image retrieval system. In the figure, 21 is an inference means for inferring an input sentence, 22 is a skeleton description means for describing an image, and 23 is an inference means 21. 24 is an image file that stores the target image;
Reference numeral 5 denotes an image description means for storing the contents of the corresponding image file 24.

The conventional image search system configured as described above operates as follows.

Various images are recorded in the image file means 24 in advance. Then, when a user wants to obtain an image that corresponds to an image search request such as "an image with a person in it,""an image of a house with square windows," or "an image on a sunny day," the user can input the following: If you input ``r person'' from the keyboard, files with the heading ``person'' will be searched. Furthermore, when "house", "window", and "square" are entered, the tree structure described in the image description means 25 is traced, and an image corresponding to "house with square windows" is retrieved.

Problems to be Solved by the Invention However, with the above configuration, it is not possible to generate an image for an arbitrarily given natural language, and the target image must be stored in an image file in advance. be. As a result, it is necessary to keep the storage capacity of the image file 24 enormous.

In view of the problems of the conventional image search system, the present invention
It is an object of the present invention to provide an image processing device that freely generates an image corresponding to the content of a natural language character string by inputting it.

Means for Solving the Problems The present invention provides a syntactic analysis means for analyzing the syntax of an input character string, a contour storage means for storing the contour of one image, and a contour attribute dictionary for storing the content and operation of the contour. an image attribute analysis unit that obtains a corresponding contour by using the contour attribute dictionary and the contour storage unit according to the result of the syntax analysis unit; and an image dictionary that stores images;
According to an image attribute dictionary that stores the content and attributes of the image, and the results of the image attribute analysis means, an image corresponding to the obtained contour is mapped using the image dictionary and the image attribute dictionary. The image processing apparatus includes a mapping means for synthesizing images.

Effect The present invention uses a syntactic analysis means to perform the following on an input string:
The syntax is analyzed, the syntax is analyzed by the image attribute analysis means, the corresponding contour is obtained using the contour attribute dictionary and the contour storage means, and the mapping means is used according to the result of the image attribute analysis means. , an image dictionary and an image attribute dictionary are used to map the obtained contour to the corresponding image to synthesize the image.

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 shows a block diagram of an image processing apparatus in an embodiment of the present invention. In the figure, 23 is a syntactic analysis means for performing syntactic analysis on an input sentence, and is a means for performing morphological analysis and determining the content of the sentence by matching with the language morphological dictionary (2). . Reference numeral 4 denotes a contour dictionary for storing the contours of images, and those of the language morpheme dictionary 3 having the necessary contours are stored in the form of patterns. 5 is a contour attribute dictionary, a contour dictionary! 4, the content attributes of the contours are stored. The image attribute analysis means 6
According to the result of the syntax analysis means 23, the contour attribute word 1]
4 and the contour storage means 5 to obtain a corresponding contour.

8 is an image dictionary that stores all images corresponding to morphemes. Reference numeral 7 denotes an image attribute dictionary that stores information about the contents and attributes of the images in the image dictionary 8. Reference numeral 9 denotes mapping means, which maps the contents of the image dictionary 8 to the contour by changing the contents address of the video memory means 1o according to the contour extracted shape.

The operation of the embodiment of the image processing apparatus configured as described above will be described below.

As an example, consider the case where the following sentence is input.

``Two men are walking in the mountains.'' The language morpheme dictionary 3 stores independent words and dependent words extracted from the Japanese dictionary. In this, there is a ``mountain'',
Contains independent words such as ``ni'', ``two people'', ``r man'', and ``walk'', as well as adjunct words ``te'', ``ni'', ``no'', ``wo'', and ``ga'', which are independent words. The input sentences are compared in order from the word dictionary and the matched sentences are considered as the analysis results. If multiple words are matched, the longest independent word is used as the matching result.

Next, the parts from which independent words are omitted are matched using an adjunct word dictionary to determine adjunct words. Next, a check is made for the matched adjuncts and independent words to see if they can be combined with the adjunct attributes. As a result, if a connectable pair of an independent word and an adjunct word can be determined, the syntactic analysis is finished,
If it cannot be determined, the same operation is performed on the second candidate adjunct word that has been matched.

As a result, ``mountain'', ``no'', ``chu'', ``two people'', ``
The results are "man", "walking", and "there".

The image attribute analysis means 6 compares these characters with the characters stored in the contour attribute dictionary 4, and uses known computer graphics techniques to determine whether a person's movements are ``r-man'' or ``walk''. The walking outline is output to mapping means 9.

The mapping means 9 then searches the image attribute dictionary 7,
Take out ``mountain'' and ``man.'' A plurality of images corresponding to unique words are stored in the image dictionary 8, and among them, images with high priority are outputted in a predetermined order. Then, the obtained contour (for example, the contour of a walking motion)
The corresponding image (for example, a man) is mapped using computer graphics technology to synthesize the image. The video memory means IO is then used for changing the address of the image, etc.

In FIG. 1, reference numeral 1 denotes a pattern recognition means, which inputs handwritten text or voice and performs text recognition and voice recognition.

The voice attribute analysis means 7 is a means for analyzing voice attributes according to the output of the syntax analysis means 23, and the voice synthesis means 12 analyzes voice attributes based on the results of the voice attribute analysis means 11.
It is a means of synthesizing speech.

As described above, according to this embodiment, it is possible to arbitrarily generate a corresponding image for any input natural language, and it is also possible to output it as audio.

In this embodiment, each means can be realized in software using a computer, but is not limited to this, and can also be realized in a dedicated hardware circuit equipped with those functions.

According to the detailed description of the invention, the present invention includes a parsing means;
Since it is equipped with mapping means, from natural language sentences,
Any image can be generated, and its practical effects are great.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an image processing apparatus according to an embodiment of the present invention, and FIG. 2 is a block diagram of a conventional image search system. l... Recognition means, 23... Syntactic analysis means, 4...
Contour attribute dictionary, 5... Contour dictionary, 6... Image attribute analysis means, 7... Image attribute dictionary, 8... Image dictionary, 9
... Mapping means, 10... Video memory means. Agent Patent Attorney Matsu 1) Tadashi Michi Figure 1

Claims (3)

[Claims] (1) A syntax analysis means for analyzing the syntax of an input character string, a contour storage means for storing the contour of an image, a contour attribute dictionary for storing the content and operation of the contour, and a syntax analysis means for analyzing the syntax of the input character string. Accordingly, an image attribute analysis means for obtaining a corresponding contour by using the contour attribute dictionary and the contour storage means; an image dictionary for storing images; and an image attribute dictionary for storing contents and attributes of the images; It is characterized by comprising a mapping means for mapping an image corresponding to the obtained contour and synthesizing an image while using the image dictionary and the image attribute dictionary according to the result of the image attribute analysis means. Image processing device. (2) It has a speech attribute analysis means that analyzes the speech attributes based on the result of the syntax analysis means, and a speech synthesis means that converts this into speech based on the output thereof, and has a speech attribute analysis means that analyzes the speech attributes based on the result of the syntax analysis means, and a speech synthesis means that converts this into speech based on the output thereof, The image processing device according to claim 1, wherein the image processing device outputs images and sounds. (3) The image processing apparatus according to claim 1 or 2, wherein the input section is provided with a recognition means, and the input character string is generated by this recognition.