Canon Kabushiki Kaisha

Noriyasu Hashiguchi

Kinya Osa

Masahiro Ariizumi

Hisashi Ishikawa

Kenta Hosaki

Yusuke Yachide

Plural block images are generated by dividing an input image into blocks each of which has a predetermined size. Clustering of pixels contained in the plural block images is performed based on a representative point associated with each of those block images to apply region segmentation to those block images. The representative point is a part of plural representative points disposed in the input image. The clustering in a target block image of the region segmentation is performed by referring to a representative point in the target block image, a representative point of a processed block image, and a representative point of an unprocessed block image. The processed and unprocessed block images are adjacent to the target block image.

Image processing apparatus and image processing method

A clustering processing unit sets representative data to be used in first clustering, for each of temporally continuous images, and repeats clustering of pixels based on the representative data. A representative data updating unit updates the representative data whenever clustering is repeated. As the representative data to be used in first clustering of each image, an initial value or representative data used in last clustering of an immediately preceding image is set based on predetermined input information. An example of the predetermined input information is operation information for an image capturing unit for capturing an image. If a dead cluster predictively forms due to an image change, the initial value is set as the representative data. This makes it possible to restrict accumulation of dead clusters, and suppress the decrease in segmentation accuracy.

Image processing apparatus and image processing method of performing image segmentation

A region segmentation apparatus for segmenting an image into a plurality of regions by pixel data clustering comprises a clustering processing unit that clusters each pixel configuring the image to one of a plurality of representative data items based on respective pixel data. Based on pixel data belonging to each representative data item as a result of the clustering by the clustering processing unit, the plurality of representative data items used in the clustering are updated for a next clustering. The region segmentation apparatus further determines whether by the clustering a state in which a dead cluster should be resolved occurs, and, if it is determined that the state in which the dead cluster should be resolved occurs, changes representative data corresponding to the dead cluster to resolve the state.

Region segmentation apparatus and method

To improve a total throughput of graph generation processing. For doing so an inputted image is divided into a plurality of areas by clustering, based on a plurality of representative points, a plurality of pixels in the image. A set of representative points that are close is generated based on positions of the representative points. A plurality of the areas corresponding to the plurality of representative points included in the set based on respective feature amounts is unified.

Image processing apparatus and method of controlling the same

An edge graph in which a pixel of an image is set as a node and an edge is set between nodes is generated. The dissimilarity or similarity between nodes at the two ends of the edge is used as the feature amount of the edge, and the edge is classified into one of a plurality of classes based on the feature amount. The edge of interest is selected in ascending class order of the feature amount, and it is determined whether to merge determination target regions to which the nodes at the two ends of the edge of interest belong. Determination target regions determined to be able to be merged are merged, and a feature amount in the merged region is updated.

The present invention enables a pattern identifying apparatus that calculates a feature amount and identifies a predetermined pattern, such as a face, based on the calculated feature amount to perform processing for reading a large volume of data at a high speed. To achieve this, a coprime relationship is established between an interval between adjoining processing windows arranged in an image and the number of memories in which the image is interleaved and stored, thereby always establishing an exclusive relationship between the memories from which data at the same position relative to reference points in the respective processing windows is read. It is thus possible to read data simultaneously, resulting in achievement of speedup.

Pattern identifying apparatus

An information processing apparatus includes an information input section that reads input image information, a cumulative image information generator that generates cumulative image information for pixels corresponding to positions of pixels of a prescribed pixel pattern in the read input image information, and a memory controller that stores in the cumulative image information holding memory the cumulative image information generated in the cumulative image information generator. This configuration allows memory resources necessary for holding the cumulative image information to be reduced.

Information processing apparatus and method for classifier-based object detection

A pattern identifying apparatus comprises: a generating unit to generate, from input data corresponding to an area, accumulated information corresponding to each position in the area; plural storing units to hold the accumulated information for each dimension; a writing unit to write the accumulated information corresponding to each position to one of the plural storing units according to a predetermined rule concerning the corresponding position; a parallel reading unit to read the accumulated information in parallel from the plural storing units; a feature quantity calculating unit to calculate a feature quantity of a local area by using the read accumulated information; and an identifying unit to identify a predetermined pattern by using the plural feature quantities. The number of the storing units allocated to at least one dimension of the area corresponding to the input data and a reading interval of the accumulated information are in a coprime relation.

Pattern identifying apparatus, pattern identifying method and program

A pattern identification apparatus computes, for each of multiple data elements in input information stored in an array, an accumulated information value corresponding to the position of the data element in the array, writes each of the computed accumulated information values into one of multiple storage units provided in advance, based on the positions of the corresponding data elements in the array, reads out, in parallel and one each from the storage units, the accumulated information values decided based on the positions in the array that corresponds to a local area, the local area being an area within an area of interest in the input information, calculates a feature amount of the local area using the accumulated information values read out from the storage units, and determines whether or not a predetermined pattern is present in the area of interest by comparing the calculated feature amount with pre-set dictionary data.

Pattern identification apparatus, control method and program thereof

An image processing method includes sending test image data to a plurality of image recognition units configured to detect a recognition object from an image, setting an evaluation condition for evaluating a recognition result, evaluating a recognition result of the test image data by each of the plurality of image recognition units under the evaluation condition, and selecting from the plurality of image recognition units an image recognition unit to be used based on an evaluation result by the evaluation.

Image processing apparatus and image processing method for evaluating a plurality of image recognition processing units

An image processing apparatus which is capable of suppressing an increase in the circuit size of buffers between data-processing circuits, thereby enabling an associated component thereof to be implemented by hardware. A position control unit sequentially shifts a position of a sub window image by a predetermined skip amount in a predetermined scanning direction, for scanning, and further repeating the scanning for skipped sub window images, after shifting a start position of the scanning, to thereby determine positions of all sub window images each as an area from a face image is to be detected.

Image processing apparatus and method

This invention scans a block using a window having a size of p&#xd7;q pixels, and encodes p&#xd7;q sub-block data each formed from pixel data at relatively identical positions in windows. A blocking unit receives a 16&#xd7;16-pixel block image. A sub-block division unit scans the received block for each 2&#xd7;2-pixel window, and generates 2&#xd7;2 sub-block data each having a size of 8&#xd7;8 pixels. Each sub-block is formed from pixel data at relatively identical positions in windows. A variable length coding unit encodes each sub-block data. When the sum of encoded sub-block data exceeds an allowable encoded data amount, a sub-block elimination unit eliminates encoded sub-block data until the sum becomes equal to or smaller than the allowable encoded data amount. The sub-block elimination unit outputs the remaining encoded sub-block data as encoded data of a block of interest.

Growing community of inventors

Kawasaki, Japan

Noriyasu Hashiguchi