Abstract: In the image recognition processing, input color image data is analyzed, and a specific pattern contained in the graphics of the bill, security or the like is detected. The color image data is binarized with regard to specific colors. In order to detect the specific pattern from the binary image, a filter is sequentially moved to scan the entirety of the binarized image data. The examination target image within the filter is subjected to linear conversion (without changing the size thereof) at each examination position, and the degree of correlation between the pre-linear conversion image and the post-linear conversion image are calculated. It is then determined based on the calculated degree of similarity (i.e., the degree of matching) whether or not the specific pattern is included in the examination target image.

Abstract: In order to retrieve images similar to a key image out of objective images, feature values of the key image and feature values of each objective image are obtained. On the basis of the feature values of the key image, weighting factors corresponding to types of the feature values are obtained. From the feature values of the key image and the each objective image, similarity values corresponding to types of the feature values are obtained. Further, a total similarity value is obtained on the basis of the similarity values and the weighting factors. The images similar to the key image are retrieved in accordance with the total similarity values of the objective images.

Abstract: An image data processing apparatus for extracting a basic texture pattern as an identifiable feature of an image so that accurate image search can be performed by comparing images using the extracted basic texture patterns. The image data processing apparatus inputs image, detects textural periodicity in the image, extracts the basic texture pattern based on the calculated periodicity, normalizes the extracted basic pattern and shifts the normalized basic pattern. The apparatus stores the normalized and shifted basic pattern to an image database with the image from which the basic pattern was extracted.

Abstract: In an image processor for detecting a circular pattern in an image, input image data is binarized to provide bi-level image data, and pixels having a predetermined value is counted in a block of a polygon having n vertices in the bi-level image data, wherein n denotes a natural number equal to or larger than eight. Then, it is decided, based on a number of the pixels having the predetermined value counted by the counter, whether the circular pattern is detected in the image or not. A detection window is used to detect the specified pattern. The detection window is moved successively by a predetermined number of pixels, in a direction from a side towards the center. The moving distance and direction are controlled based on the result of scan.

Abstract: In an image processor, it is decided that a target pixel has a specified color when the input color data of the pixel are decided to exist in the first ranges and the differences between color data of the pixel and those of pixels adjacent thereto are decided to exist in the second ranges. Thus, the specified color in a specified pattern can be detected even when the input image data are affected by external factors. Alternatively, it is decided that the pixel has a specified color when the input color data of the pixel are decided to exist in the first ranges and the results of calculation on the input color data are decided to exist in the second ranges. Errors in color detection can be decreased by narrowing the color detection ranges with use of different types of conditions.

Abstract: An image processor detects a specified pattern defined by arrangement of three or more elements. Positions of a plurality of element candidates are detected based on input image data, and n pairs (n≧2) of element candidates are selected in the plurality of elements. Then, n arrangement criterions are determined based on the positions of the selected n pairs, and one arrangement criterion is determined based on the n arrangement criterions. By using the one arrangement criterion, positions of all the elements are calculated in the specified pattern. The specified pattern is detected based on the calculated positions of the arrangement.

Abstract: An image data processing apparatus for extracting a basic texture pattern as an identifiable feature of an image so that accurate image search can be performed by comparing images using the extracted basic texture patterns. The image data processing apparatus inputs image, detects textural periodicity in the image, extracts the basic texture pattern based on the calculated periodicity, normalizes the extracted basic pattern and shifts the normalized basic pattern. The apparatus stores the normalized and shifted basic pattern to an image database with the image from which the basic pattern was extracted.

Abstract: A similar-image retrieving apparatus and a similar-image retrieving method allow data amount to be reduced by digitizing feature quantities of external shapes of islands extracted from images, and also allow shape comparison to be achieved by simple calculation of the digitized feature quantities. Thus, the apparatus and the method are capable of practical retrieval of similar images. A feature quantity extracting section 2 extracts an external shape of an island from image data, and determines coefficients by Fourier series expansion of point strings of edge pixels of the external shape, thereby extracting the coefficients as feature quantities of the external shape of the island. Based on the feature quantities of a key image and the feature quantities of comparative images extracted in this way by the feature quantity extracting section 2, similarities between the key image and the comparative images are calculated by a similarity calculating section 6.

Abstract: In image data retrieval, image data that becomes the base of search is specified. When the search is to be carried out according to the distance from the image data on an image map, an automatic setting process of retrieval information is carried out. Here, a keyword, icon, chromaticity and the like of that image data are automatically set as retrieval information. The position of that image data on the image map becomes the center of a range when the ambiguity is set. By setting the ambiguity of the image map, keyword, icon, chromaticity, and the like in this state, the range of ambiguity is displayed, and image data within that range is automatically retrieved. As a result, analogous image data can be searched for according to subtle and ambiguous information or information that cannot be quantified.