Abstract: A production unit of an image processing apparatus produces a contour signal of an image including a specific subject. A detection unit detects, on the basis of the contour signal, a representative contour direction for each of a plurality of division regions of the image, where the detection unit detects a specific direction as the representative contour direction when the direction of the entire contour portion included in the division regions is biased in the specific direction by at least a specific degree. A determination unit determines a type of the subject on the basis of at least one of a direction-based frequency distribution of the detected representative contour directions, a number of representative contour directions detected, etc. A correction unit configured to correct the image data according to a correction method corresponding to the type of the subject.

Abstract: A face-tracking method with high accuracy is provided. The face-tracking method includes generating an initial face shape according to the detected face region of an input image and a learned data base, wherein the initial face shape comprises an initial inner shape and an initial outer shape; generating a refined inner shape by refining the initial inner shape according to the input image and the learned data base; and generating a refined outer shape by searching an edge of the refined outer shape from the initial outer shape toward the limit of outer shape.

Abstract: Disclosed herein is a method for auto-depicting trends in object contours (referred to as ADTOC). At the heart of ADTOC is a sifting process to determine a significant angular value via evaluating a plurality of angular values in a predefined range. ADTOC is characterized in that a probe-ahead concept is applied to obtain a reference angular value along the current route, and then the probed angular value is used to modify the significant angular value in order to timely correct the subsequent trace direction, thus achieving more accurate trace result. Contours with discontinuous segments caused by noise, obstacles, illumination, shading variations, etc. can also be auto-depicted without requiring a predefined auxiliary route.

Abstract: A computer-implemented method and apparatus for detecting an object of interest. An edge image is generated from an image of a scene. A sectioned structure comprising a plurality of sections is generated for use in analyzing the edge image. The edge image is analyzed using the sectioned structure to detect a presence of the object of interest in the edge image.

Abstract: A video image capture component includes a light source operable in a first spectrum, a first image detector operable in the first spectrum, a second light source operable in a second spectrum, and a second image detector operable in the second spectrum. A filtering component generates a combination image by filtering a first image obtained by the first image detector with a high-contrast filter, resulting in a high-contrast image, and masking a second image obtained by the second image detector using the high-contrast image. A compositing component creates a composite image from the combination image and a selected image. A display component displays the composite image. Alternative systems and methods for creating a combination image include techniques involving thermal imaging, laser detection, and narrow band frequency detection.

Abstract: Embodiments are provided for content item classification.

Abstract: Disclosed are an image processing device, an image processing method, and a non-transitory computer-readable medium that obtain a contour image in which noise edges are eliminated from local edge information and a contour of an important object is enhanced. The image processing device includes a local contour extraction unit which generates a local edge image and a global contour extraction unit which generates a global edge image. The image processing device generates a contour image by preparing the local edge image and obtaining the weighting sum of the local edge image and the global edge image.

Abstract: An apparatus for detecting an afterimage candidate region includes: a comparison unit which compares gradation data of an n-th frame with integrated gradation data of an (n?1)-th frame and generates integrated gradation data of the n-th frame, where n is a natural number; a memory which provides the integrated gradation data of the (n?1)-th frame to the comparison unit and stores the integrated gradation data of the n-th frame; and an afterimage candidate region detection unit which detects an afterimage candidate region based on the integrated gradation data of the n-th frame, where each of the integrated gradation data of the n-th frame and the integrated gradation data of the (n?1)-th frame comprises a comparison region and a gradation region.

Abstract: Methods and systems described herein detect object boundaries of videos. A window around the pixel may be followed in adjacent image frames of the image frame to determine object boundaries. Inconsistencies in image patches over a temporal window are detected, and each pixel of the image frame of a video is assigned an object boundary probability. The pixel may belong to a texture edge if the window content does not change throughout the adjacent image frames, or the pixel may belong to an object boundary if the window content changes. A probability value indicating the likelihood of the pixel belonging to an object boundary is determined based on the window content change and is assigned to the corresponding pixel.

Abstract: A method, apparatus, system, and computer program product provide the ability to model a polyline boundary from point cloud data. Point cloud data is obtained and boundary cells are extracted. Potential boundary points are filtered from the boundary cells. Line segments are extracted from the potential boundary points and refined. A regularized polygon is obtained by intersecting the refined line segments.

Abstract: In one aspect, there is disclosed a digital signal processor and method performed by the same for performing object detection, including facial detection, in a reduced number of clock cycles. The method comprises using Sobel edge detection to identify regions with many edges, and classifying those regions as foreground candidates. Foreground candidates are further checked for vertical or horizontal symmetry, and symmetrical windows are classified as face candidates. Viola-Jones type facial detection is then performed only on those windows identified as face candidates.

Abstract: Techniques described herein convert digital images into images suitable for use as coloring book pages and the like. The techniques include applying a smoothing operator to at least one channel image of a digital input image to form at least one smoothed image. A scaling operator is then applied to the at least one smoothed image to form at least one scaled image. A tone equalization operator is then applied to the at least one scaled image to form at least one equalized image. An edge detection operator is then applied to the at least one scaled image to form at least one edge image. In addition, a threshold operator is applied to the at least one edge image to form at least one threshold image. A digital line drawing is then created based at least in part on the at least one threshold image.

Abstract: A method and system for identifying an edge in a portion of an image, the portion having an associated intensity variation. An edge candidate is detected at a local extremum of the intensity variation (i.e. at a position along the x-axis that corresponds to a peak or valley of the intensity variation). A local threshold associated with the edge candidate is computed on the basis of a slope tolerance and the slope value of the edge candidate, where this local threshold defines a boundary region for the edge candidate. If the edge candidate is determined to be a global extremum (i.e. the highest peak or lowest valley) within its respective boundary region, the edge candidate is verified as being an actual edge of the image portion. Identification of this verified edge may include determination of one or more of a plurality of characteristics of the edge.

Abstract: A computer system may elicit from a human observer ground truth data useful in automatically detecting one or more features in images. The elicitation may include presenting an image to a human observer that has a visual indicator in an image, the visual indicator indicating having a location and orientation with respect to the image; asking the human observer to judge whether a particular image feature is present in the image at the location and orientation indicated by the visual indicator; receiving input from the human observer indicative of whether the particular image feature is present at the location and orientation indicated by the visual indicator; storing the input received from the human observer as part of the human-labeled ground truth data; and repeating the process described above one or more times in connection with a visual indicator that has a different location or orientation with respect to the image or that uses a different image.

Abstract: A mask correcting unit corrects an externally set mask pattern. A depth map processing unit processes a depth map of an input image for each of a plurality of regions designated by a plurality of mask patterns corrected by the mask correcting unit. An image generation unit generates an image of a different viewpoint on the basis of the input image and depth maps processed by the depth map processing unit. The mask correcting unit performs a blurring process on an object boundary part of the mask pattern.

Abstract: An image processing apparatus capable of accurately determining an inclination of an original even in the case where an original of a white sheet is read with a white sheet as a background is provided. For this, the image processing apparatus extracts a fiber section having a feature of paper fiber from image data, analyzes a direction of fiber in the fiber section, and calculates an inclination of an original at the time of reading the original based on the direction.

Abstract: An apparatus may include a motion sensing device, a communication interface; and a processor. The processor may receive a point of origin that corresponds to a spatial position of the apparatus. The processor may determine, responsive to detecting a user interface command and based on a plurality of values received from the motion sensing device, a spatial position of the apparatus relative to the point of origin, and to acquire an image of an object in a field of view of the apparatus. The processor may also determine a scale factor of the image based on at least the spatial position of the apparatus relative to the point of origin.

Abstract: With this device, an edge pair likelihood map generation unit (120) computes an edge pair likelihood which denotes the plausibility that a pixel of a pair is an edge of an estimation subject site, and generates an edge pair likelihood map which denotes the edge pair likelihood for each pixel. A continuity likelihood map generation unit (130) evaluates, upon the edge pair likelihood map, the continuity for the edge pair likelihood of a pixel which is included in a region wherein the estimate subject site is presumed, and generates a continuity likelihood map which denotes an edge pair likelihood which has continuity as a candidate region of the estimate subject site. An integrated likelihood map generation unit (140) generates an integrated likelihood map which denotes the candidate region which the continuity likelihood map denotes, refined based on a predetermined condition.

Abstract: A method is disclosed to automatically segment 3D and higher-dimensional images into two subsets without user intervention, with no topological restriction on the solution, and in such a way that the solution is an optimal in a precisely defined optimization criterion, including an exactly defined degree of smoothness. A minimum-cut algorithm is used on a graph devised so that the optimization criterion translates into the minimization of the graph cut.

Abstract: A method for tracing edges of an image using hysteresis thresholding includes: (i) receiving an edge map of the image, (ii) scanning one row of the input edge map, (iii) assigning a label to each edge pixel in the row based at least in part on the presence or absence of a neighboring edge pixel, (iv) grouping contiguous labels, and (v) identifying groups of edge pixels.

Abstract: An image processing apparatus is disclosed, including a modification part, a model creation part, and a division part. The modification part instructs modification of a portion of a foreground area or a background area which is erroneously divided as a result of dividing predetermined image data into the foreground area being an extraction target and the background area being an out-of-extraction target. The model creation part creates a foreground model of the foreground area and a background model of the background area which are modified, based on first pixel values of a predetermined line segment where the modification is instructed and second pixel values of the foreground area and the background area which are divided. The division part divides the image data into the foreground area and the background area by using the foreground model and the background model which are created by the model creation part.

Abstract: A surface definition module of a hair/fur pipeline may be used to generate a shape defining a surface and an associated volume. A control hair module may be used to fill the volume with control hairs and an interpolation module may be used to interpolate final hair strands from the control hairs.

Abstract: A method for determining a relatedness between a query video and a database video is provided. A processor extracts an audio stream from the query video to produce a query audio stream, extracts an audio stream from the database video to produce a database audio stream, produces a first-sized snippet from the query audio stream, and produces a first-sized snippet from the database audio stream. An estimation is made of a first most probable sequence of latent evidence probability vectors generating the first-sized audio snippet of the query audio stream. An estimation is made of a second most probable sequence of latent evidence probability vectors generating the first-sized audio snippet of the database audio stream. A similarity is measured between the first sequence and the second sequence producing a score of relatedness between the two snippets. Finally a relatedness is determined between the query video and a database video.

Abstract: A method for recognition of a predetermined pattern in an image data set recorded by a device for recording of at least two electromagnetic frequency spectra is provided. A first difference value is formed for the image points of the selected area as a function of a difference between a data vector of a corresponding image point and a first reference data vector. A second difference value is formed for an image point of a selected area as a function of a difference between the data vector of this image point and a second reference data vector. A predetermined pattern is recognized when it is determined at least one pattern correlation quantity is below a predetermined threshold value and a local minimum is present.

Abstract: Implementations relate to estimating noise in images. In some implementations, a method includes extracting a plurality of sample blocks of pixels from a received image, where each sample block includes a subset of pixels of the image. One or more of the sample blocks are examined for texture content based on whether the sample blocks include one or more edges based on a predetermined threshold. At least one sample block determined to include texture content is removed. The method determines one or more average color variances based on the remaining sample blocks that have not been removed, where noise estimations for the image are based on the average color variances.

Abstract: A method, system, apparatus, article of manufacture, and computer readable storage medium provide the ability to fill pixels of an image. An area of the image is obtained and one of the pixels is identified as an anchor pixel. Waves are created based on the anchor pixel and placed into a collection of waves waiting to be processed. Each of the waves in the collection is processed. A type of pixel in front of a current wave being processed is determined. If the pixel is fillable, the pixel is filled, the wave is advanced, and child waves are updated. If the pixel is blocking the advancement of the wave, the current wave is updated (e.g., via shrinking or splitting), and child waves are updated. Once all waves have been processed, the area of the image is displayed based on the fill.

Abstract: An object of the present invention is to provide an edge detection technique and equipment which are capable of stably detecting an edge by suppressing the influence of noise even in the case where the image is obtained by charged particle radiation equipment, such as a scanning electron microscope and has a low S/N ratio. More specifically, the present invention is to propose a technique and equipment which are configured to determine a peak position (edge) on the basis of the following two edge extraction techniques. That is, the present invention is to propose a technique and equipment wherein at least two peaks are formed by using, as edge detection techniques, for example, one peak detection technique having a relatively high sensitivity and the other peak detection technique which is relatively less susceptible to the influence of noise than the one peak detection technique, and wherein a position where the peaks coincide with each other is determined as a true peak position (edge position).

Abstract: When there are a plurality of detection candidate objects, to detect a target object that is appropriately visible as a whole, an information processing apparatus calculates detection likelihoods in a plurality of local areas of each of the plurality of detection target candidates, and a detection reliability of each of the detection target candidates based on a distribution of the detection likelihoods.

Abstract: A three-dimensional (3D) image display device may display a perceived 3D image. A location tracking unit may determine a viewing distance from a screen to a viewer. An image processing unit may calculate a 3D image pixel period based on the determined viewing distance, may determine a color of at least one of pixels and sub-pixels displaying the 3D image based on the calculated 3D image pixel period, and may control the 3D image to be displayed based on the determined color.

Abstract: A computer-implemented segmentation method is used to process an image representing a plurality of nuclei. The method is implemented in a computer having a processor and a physical memory. A set of instructions are provided to the processor the physical memory of the computer. The processor is configured by executing the set of instructions in the physical memory so as to automatically segment the image by: thresholding a grey-scale image to create a black and white image; identifying objects in the black and white image and removing objects failing to meet predetermined criteria; extracting objects; and applying an edge detector on the segmented image to identify the edges of the nuclei. Overlapping nuclei are split to improve results.

Abstract: A method of operation of a display system includes: receiving an input image having input pixels; identifying input edge pixels from the input pixels; calculating an edge pixel sum for one of the input edge pixels based on the number of the input edge pixels surrounding one of the input edge pixels; identifying pruned edge pixels from the input edge pixels having the edge pixel sum between a high edge density threshold and a low edge density threshold; and selecting a dominant image line from candidate lines calculated between the pruned edge pixels to generate a processed image for displaying on a device.

Abstract: A weighted image enhancement method includes receiving an original image. The original image includes several original pixels. The original image is sharpened to generate a sharpened image. The sharpened image includes several sharpened pixels. Edge detection is performed with respect to the original image to generate a probability of whether each original pixel is on an edge. An enhancement mode setting is received. A corresponding weight table corresponding to the enhancement mode setting is looked up to obtain a corresponding enhancement weight of each original pixel according to the probability of whether each original pixel is on an edge. A weight calculation is performed utilizing each original pixel and its corresponding sharpened pixel according to its corresponding enhancement weight to generate an enhanced image. The enhanced image is displayed by a display unit.

Abstract: According to an aspect of the present invention, there is provided a pattern matching method of detecting an image of a detection target from a search image, comprising: obtaining a reference image of the detection target; generating the model edge image on a basis of the reference image; generating the edge extraction domain that is specified as a portion where the model edge image can exist by overlying a plurality of the model edge images obtained with at least one of rotating the model edge image within a predetermined range around a rotation center of the model edge image and translating the model edge image within a predetermined range; and performing pattern matching between the model edge image and the search edge image.

Abstract: Techniques are described for controlling communication devices using image analysis. For instance, when a communication is received by a communication device, the communication device outputs an alert to notify users of the received communication and accesses one or more images of an area proximate to the communication device. The one or more images cover an area proximate to the communication device at a time during which the communication device is outputting the alert. The communication device analyzes the one or more images to determine whether a user is present in the one or more images and in a position to perceive the received communication. The communication device handles at least one aspect of the received communication based on the determination of whether a user is in a position to perceive the received communication.

Abstract: The invention provides methods and systems for reconstructing feature intensities from pixel level data. In certain embodiments, the invention uses an empirically determined transfer function to construct a theoretical estimate of pixel level data and then iteratively updates feature intensities based on a minimum multiplicative error between the pixel level data and the theoretical estimate of the pixel level data.

Abstract: An activity recognition method, for recognizing continuous activities of several moving objects in the foreground of a video, includes: capturing and processing a training video to get a contour of a moving object; extracting a minimum bounding box of the contour in order to get parameters then transfer to feature vectors; constructing a decision tree model based on support vector machines (SVMs), for classifying the activities of the moving object according to the parameter and the feature vector of the training video; capturing and processing a testing video to get other parameters and using several formulas to generate feature vectors, and executing an algorithm for recognizing the activities of several moving objects in the foreground of the testing video. Said feature vectors are transformed from the parameters that in the testing and training videos. Via above descriptions, the method can recognize activities of foreground objects in the testing video.

Abstract: An image processing device includes a processor, and a memory storing computer-readable instructions therein. The computer-readable instructions, when executed by the processor, causes the image processing device to perform: generating edge image data by using the original image data; calculating characteristic values for a plurality of determination regions; and identifying a determination region as a nonuniform region when the characteristic value of the determination region satisfies a prescribed criterion, and the determination region as a uniform region when the characteristic value of the determination region does not satisfy the prescribed criterion. Each of the plurality of determination regions corresponds to one of the characteristic values, represents a part of the edge image, and includes a plurality of pixels, the plurality of determination regions being different from one another, each of the characteristic values characterizing the edge strength of the corresponding determination region.

Abstract: Image restoration by restoring a high-frequency component of a deteriorated image. A high-frequency image information predicting and stabilizing method includes: determining a shooting area, which is a high-frequency area that includes an overly-restored high-frequency component, in an input image; predicting a high-frequency component of the input image by applying different deterioration models to the shooting area and a remaining area in the input image; and stabilizing a high-frequency component of an edge area by using an edge blur kernel determined to be parallel to an edge direction of the input image.

Abstract: A computing device performs matching between a target image and one or more template images. The computing device receives image data and performs an edge detection algorithm on the image data. The edge detection algorithm includes a distance metric based on angles between gradient vectors in the image data and gradient vectors in one or more templates. The computing device matches a building model to the image data based on results of the edge detection algorithm, wherein the building model is associated with the one or more templates.

Abstract: An arc detecting apparatus includes an input unit that inputs an image; a line-segment detector that detects line segments from the image; a determiner that determines whether two of the line segments are associable with each other, on a basis of positions of the two line segments and angles of the two line segments relative to corresponding references; and an arc detector that detects an arc approximated by at least two line segments including the two line segments, on a basis of the two line segments being associated with each other according to a result of the determination.

Abstract: A method and system for orientation compensation of a mobile device within an environment includes providing a plurality of reference markers having straight edges and having a defined, regular orientation with respect to the environment. Information about the orientation of the reference markers is supplied to a mobile device operating within the environment. An orientation sensor disposed within the mobile device estimates an orientation of the mobile device. An image of one reference marker is captured, and at least one edge of that reference marker is located. The estimated orientation is compensated by correcting for the reference marker orientation and aligning the corrected estimated orientation to the at least one edge of the reference marker that is closest to being parallel to the corrected estimated orientation.

Abstract: An adaptive loop filter classifies a decoded image from a deblocking filter into a smooth region class and an edge/texture region class, in accordance with results of edge detection from an edge detecting unit. The adaptive loop filter performs filter coefficient calculation for each of classified classes such that the residual difference between an original image from a screen rearranging buffer and an image from the deblocking filter is smallest. The adaptive loop filter performs filter processing using filter coefficients calculated for each classified class, and outputs an image after filter processing to frame memory. The filter may be applied to an image encoding device for performing encoding with the H.264/AVC format as a basis, for example.

Abstract: There is provided a defect inspection method capable of detecting a crack with high accuracy. The defect inspection method includes the steps of: obtaining a shot image comprising pixels; and scanning the shot image in predetermined directions, and assigning a high evaluation value to a pixel M for each scanning direction when the luminance of the pixel M is lower than the luminances of first adjacent pixels K, O, located on both sides of the pixel M in the scanning direction and, in addition, the luminance of each of second adjacent pixels C, W, located on both sides of the pixel M in a direction perpendicular to the scanning direction, is lower than the luminances of third adjacent pixels A, E or U, Y located on both sides of the second adjacent pixel in the scanning direction.

Abstract: The invention is directed to detecting a boundary position between a foot and a lower leg of a person in an image acquired by an imaging unit, the boundary position being a substantial boundary part, in a lower limb, between the foot, which is a part from a malleolus to a tip part, and the lower leg; detecting a feature quantity that makes it possible to classify a ground and a part other than the ground in the image; setting, in a peripheral region around the boundary position, a plurality of local regions having positional information and/or direction information relative to the boundary position, and determining whether each of the local regions is the ground or the part other than the ground by using the feature quantity unique to the ground; determining a foot region from the local region determined as the part other than the ground; and estimating a direction of the foot of the person from the local region classified as the foot region and from the information.

Abstract: An object identification system calculates boundaries between real objects from images of the real objects, and calculates first indicators which correspond to each angles of a first angle section divided into a first angle gap and which varies every boundary between the calculated real objects. The object identification system extracts a virtual object having an outline firstly meet with a radiating line corresponding to each map angle of the divided into a second angle gap, and generate a set of second indicators corresponding to each map angle. The object identification system matches the first indicators into second indicators having a repeat ratio substantially equal to a repeat ratio of the first indicators in an angle section, and extracts virtual objects matched with each of the previewed real objects.

Abstract: A reliable method for discriminating between a plurality of edges in a region of interest of an edge feature video tool in a machine vision system comprises determining a scan direction and an intensity gradient threshold value, and defining associated gradient prominences. The gradient threshold value may be required to fall within a maximum range that is based on certain characteristics of an intensity gradient profile derived from an image of the region of interest. Gradient prominences are defined by limits at sequential intersections between the intensity gradient profile and the edge gradient threshold. A single prominence is allowed to include gradient extrema corresponding to a plurality of respective edges. A gradient prominence-counting parameter is automatically determined that is indicative of the location of the selected edge in relation to the defined gradient prominences. The gradient prominence-counting parameter may correspond to the scan direction.

Abstract: The present invention discloses an image-based barrier detection and warning system and a method thereof, wherein an image processing technology is used to establish ROI and verify whether a horizontal line signal exists, and wherein the image processing technology determines whether an object is a barrier via detecting contours of an object and detecting distance to the object, and wherein the system timely outputs warning signals and presents distance to a barrier when detecting the barrier. The image-based barrier detection and warning system enables the driver to watch the surroundings of the vehicle directly and clearly and learn the relative positions of barriers, whereby the driver can park more easily and safely.

Abstract: Method, system, device and computer program product for projective correction of an image containing at least one text portion that is distorted by perspective. The method includes a step of image binarization involving binarizing said image. The method includes connected component analysis. Pixel blobs are detected in said at least one text portion of said binarized image in connected component analysis. The method includes horizontal vanishing point determination, including estimating text baselines by means of eigenpoints of said pixel blobs and determining a horizontal vanishing point of said at least one text portion by means of said text baselines. The method also includes vertical vanishing point determination. A vertical vanishing point is determined for said at least one text portion on the basis of vertical features thereof. The method includes projective correction. Perspective in said image is corrected on the basis of said horizontal and vertical vanishing points.

Abstract: An information processing apparatus includes an image acquiring unit to acquire captured image of a watching target person and a target object as a reference for a behavior, the captured image containing depth information indicating depths per pixel, a foreground area extracting unit to extract a foreground area on the basis of a difference between a foreground image and the captured image, the foreground image being set to contain a depth of the foreground and a behavior presuming unit to presume the behavior about the target object by determining whether a positional relationship between the foreground area and the target object area satisfies a predetermined condition or not on the basis of referring to the depths of the pixels in the foreground area based on the depth information, the condition being set on the assumption that the extracted foreground area is related to the behavior.

Abstract: A method for performing fast detail-preserving filtering of an input digital image includes, for each pixel in the image, calculating the difference between the selected pixel and each of its four neighboring pixels located above, left, right, and below the selected pixel, calculating a scaled weighted sum of differences between the actual pixel and its four neighbors, where for each neighboring pixel the weight is an edge-sensing function having a data-adaptive scaling parameter function, and adding the weighted sum of differences to the value of the selected pixel.