Kyocera Document Solutions Inc.

Takuya Miyamoto

Kanako Morimoto

Rui Hamabe

Kazunori Tanaka

Koji Sato

Naomichi Higashiyama

Shiro Kaneko

Akinori Matsuno

Kazuhiro Nakachi

A sheet identification device includes a light irradiation portion. When forming an identification image, the light irradiation portion projects first pattern light (P) and second pattern light (P) having different dependences on a fiber direction of a surface of a sheet on an identification region (R). The identification image is an image of the identification region (R) of the surface of the sheet on which image formation or image reading is performed, which is used for identifying the fiber direction of the surface of the sheet.

Sheet identification device and image processing apparatus capable of identifying fiber direction of sheet

An object of the present invention is to determine the cause of an image defect that occurs in an image forming device with high accuracy while suppressing the amount of calculation at a processor. A processor () determines the cause of an image defect on the basis of a test image read from an output sheet of an image forming device (). The processor () generates a plurality of feature images by executing a feature extraction process on the test image. Further, the processor () uses each of the plurality of feature images as an input image and determines which of a plurality of cause candidates corresponding to the image defect the input image corresponds to by pattern recognition of the input image.

Image processing method and image processing apparatus

A feature-amount extraction unit generates a base feature-map group constituted by a plurality of base feature maps from an input image, applies a plurality of statistic calculations to the base feature maps in the base feature-map group, and generates a plurality of types of statistic maps. The inference unit derives inference results of segmentation for inference inputs based on the plurality of statistic maps. Each of the plurality of types of statistic calculations described above is processing of calculating a statistic with a specific window size and a specific calculation formula, and the plurality of types of statistic calculations are different from each other in at least either one of the window size and the calculation formula.

Image recognition method, image recognition apparatus and computer-readable non-transitory recording medium storing image recognition program

An image processing apparatus compares a target image and a reference image and thereby detects an anomaly in the target image, and includes an anomaly detecting unit. The anomaly detecting unit is configured to (a) generate a first characteristic map obtained by performing a filter process for the target image and a second characteristic map obtained by performing the filter process for the reference image, (b) derive a correction amount on the basis of a deviation between an object in the first characteristic map and an object in the second characteristic map, and (c) correct the target image and/or the reference image with the correction amount and thereafter compare the target image and the reference image and thereby detect an anomaly in the target image.

Image processing apparatus

An image recognition method includes a feature amount extracting step of generating, from an input image, a base feature map group including a plurality of base feature maps; an inferring step of deriving a plurality of inference results using each of a plurality of machine-learned inference devices for a plurality of inference inputs based on the base feature map group; and an integrating step of integrating the plurality of inference results by a specific manner to derive a final inference result, where each of the plurality of inference inputs has some or all base feature maps of the plurality of base feature maps, and each of the plurality of inference inputs has the some or all base feature maps that are different in part or whole from the some or all base feature maps of another inference input in the plurality of inference inputs.

Image recognition method using a plurality of machine-learned inference devices, image recognition apparatus for the same, and non-transitory computer readable recording medium storing a program of the same

A processor determines a cause of an image defect based on a test image that is obtained through an image reading process performed on an output sheet output from an image forming device. The processor extracts a vertical stripe part extending along a sub scanning direction in the test image. Furthermore, the processor determines which of predetermined two types of cause candidates is a cause of the vertical stripe part, based on a distribution of a pixel value sequence along a main scanning direction that crosses the sub scanning direction, in a target part including the vertical stripe part in the test image.

Image processing method, image processing apparatus

A processor determines an image defect in a test image obtained through an image reading process performed on an output sheet output by an image forming apparatus. The processor creates an image to be processed including a horizontal line, extending in a horizontal direction, extracted from the test image. Furthermore, the processor determines the presence or absence of at least one periodicity set in advance in a vertical direction in the image to be processed and determines the cause of the horizontal line according to the determination result on the periodicity.

A processor determines a cause of an image defect based on a test image that is obtained through an image reading process performed on an output sheet output from an image forming device. The processor generates an extraction image by extracting, from the test image, a noise point that is a type of image defect. Furthermore, the processor determines a cause of the noise point by using at least one of, in the extraction image: an edge strength of the noise point; a degree of flatness of the noise point; and a pixel value distribution of a transverse pixel sequence that is a pixel sequence traversing the noise point.

A processor determines an image defect in a test image obtained through an image reading process performed on an output sheet output by an image forming apparatus. The processor derives a vertical data string that is composed of data about horizontal representative values serving as representative values of a plurality of pixel values in each line in a horizontal direction in the test image. Moreover, the processor determines the presence or absence of at least one periodicity set in advance in the vertical data string by performing frequency analysis on the vertical data string. Furthermore, the processor determines the occurrence or non-occurrence and the cause of periodic density unevenness serving as a type of the image defect according to the determination result on the periodicity.

Method for performing image processing for determining image defect in mixed-color test image through image reading

A processor identifies an original drawing part that is originally drawn in a test image. Furthermore, the processor identifies a noise point in the test image. Furthermore, the processor determines which of predetermined two types of cause candidates is a cause of the noise point by determining a degree of overlapping between the original drawing part and the noise point. The two types of cause candidates are: waste toner dropping in which waste toner that has adhered to a transfer body that transfers a toner image to a sheet, is transferred to the sheet in the image forming device; and carrier developing in which magnetic carrier that has been mixed with toner is transferred to a sheet.

A vector generation unit derives a reference feature vector and a document feature vector. A feature quantity extraction unit performs a dimensionality reduction process on the reference feature vector and the document feature vector so as to set a dimensional value as a first feature quantity and derives a cosine similarity between the reference feature vector and the document feature vector as a second feature quantity. A grid division unit classifies documents into first partial regions obtained by dividing a feature quantity space of the first feature quantity, and classifies the documents into second partial regions obtained by dividing a range of the second feature quantity. A training data extraction unit selects, for each combination of a first partial region and a second partial region, a document classified in both the partial regions and sets documents selected with respect to all combinations as training data.

Training data collection system, similarity score calculation system, document retrieval system, and non-transitory computer readable recording medium storing training data collection program

A processor selects a target sheet from a plurality of predetermined sheet candidates in accordance with selection information that is input via an input device. Furthermore, the processor derives feature information regarding a noise point from a target image that is obtained through an image reading process performed on an output sheet output from an image forming device, the noise point being a dot-like noise image included in the target image. Furthermore, the processor determines whether or not the noise point is a dot-like sheet noise by applying the feature information to a determination algorithm that corresponds to the target sheet, the sheet noise being included in a sheet of the output sheet itself, the determination algorithm being one of a plurality of determination algorithms that respectively correspond to the plurality of sheet candidates.

A processor generates an extraction image by extracting, from a test image, a target specific part that is a horizontal stripe extending in a horizontal direction or a noise point. The processor derives a vertical data sequence from a target area in the extraction image and converts the vertical data sequence to primary conversion data of a frequency domain. The processor converts, to secondary conversion data of a space domain, correction data that is obtained by removing, from the primary conversion data, data of frequency bands other than a particular frequency band corresponding to a target specific part that was determined as having periodicities. The processor classifies the target specific parts into periodic specific part and non-periodic specific part by comparing positions of the target specific parts with a peak position in a waveform represented by the secondary conversion data.

The image processing apparatus includes: an abnormality detection unit configured to detect one or more abnormalities included in a target image and an abnormality exclusion processing unit configured to exclude a specific abnormality from the detected one or more abnormalities. In a case in which a detection area of a certain abnormality among the detected one or more abnormalities and a detection area of another abnormality overlap with each other and a type of the certain abnormality and a type of the another abnormality are different from each other, when brightness information of one of the certain abnormality and the another abnormality satisfies a specific condition, the abnormality exclusion processing unit excludes the one of the certain abnormality and the another abnormality from the detected one or more abnormalities.

Image processing apparatus, method, and computer program product excludes specific anomalies among detected anomalies

A processor identifies a specific part composed of a plurality of significant pixels in a mixed-color test image. Furthermore, the processor identifies a color vector. The color vector represents a vector in a color space from one of a color of the specific part in the mixed-color test image and a color of a reference area including a periphery of the specific part to the other. Furthermore, the processor determines which of a plurality of image creating portions of an image forming device corresponding to a plurality of developing colors, is a cause of an image defect, based on the color vector identified from the mixed-color test image.

Growing community of inventors

Osaka, Japan

Takuya Miyamoto