Abstract: Method and apparatus for improving the quality of super-resolution video imaging by suppressing ringing artifacts, reducing high-frequency noise, reducing blocking artifacts, and smoothing out jagged edges of the image to generate pictures that appear cleaner with less edge degradation. The method operates in a recursive manner within a sequence of low resolution images. Conventional SR processing is primarily enhanced within the invention by adding an artifact suppression section which creates a high frequency component signal ?SRi having significantly reduced artifacts therein achieving higher quality super-resolution image output. The method can be applied to images and image sequences (video) in monochrome or color and in any desired pixel format. The method can be implemented within image processing devices, in particular those containing programming for executing the described method steps.

Abstract: Simulating a long exposure-time image from a sequence of short exposure-time images captured at slightly different times. The sequence of images is combined in a temporal integration process to create a long exposure image that simulates the output from a still camera, steadied by a tripod, whose light-sensitive material has been exposed to the same scene from the time of the beginning of the first input image of the sequence to the last image of the input sequence. The method overcomes limitations of hand-held video and image recording devices, allowing the user to easily create effects normally associated with high-end digital still cameras under expert control.

Abstract: A method and apparatus is described here that categorizes images by extracting a subscene and describing the subscene with a top level feature vector and a division feature vector, which are descriptions of edge gradient classifications within rectangular bounding boxes. By filtering subscene feature vectors in images with a Gaussian mixture based model pool (obtained in a subscene modeling phase), the images may be categorized (in an subscene recognition phase) with probabilities relating to each subscene. Higher probabilities are likelier correlations. The device may be a single or multiple core CPU, or parallelized vector processor for characterizing many images. The images may be photographs, videos, or video stills, without restriction. When used real-time, the method may be used for visual searching or sorting.

Abstract: An apparatus and method for processing pictures (images, graphics or video frames) for image representation and comparison on the basis of a geometric feature description built from histograms of pseudo-color saturation. The feature description can also include normalized centroid variance, as well as an intensity map. The descriptions allow various matching comparisons to be performed between an input image and a set of comparison images, such as to find matching (or mismatching, or other relationship) images. The comparison can be sped using staged comparisons, whereby an image failing one level of comparison need not be considered in subsequent phases. A set of efficient image feature descriptors are described for use in a fast image retrieval scheme, which is efficient for searching of images spanning different image types, rotations and scales.

Abstract: A method and apparatus is described that categorizes images by extracting regions and describing the regions with a set of 15-dimensional image patch feature vectors, which are concatenations of color and texture feature vectors. By comparing the image patch feature vectors in images with similarly-obtained image patch vectors in a Gaussian mixture based model pool (obtained in an image patch modeling phase), the images may be categorized (in an image patch recognition phase) with probabilities relating to each image patch. Higher probabilities are likelier correlations. The device may be a single or multiple core CPU, or parallelized vector processor for characterizing many images. The images may be photographs, videos, or video stills, without restriction. When used real-time, the method may be used for visual searching or sorting.

Abstract: A low cost image quality enhancement method to be utilized in an ordinary consumer camera and for post processing of an existing image database is described herein. The quality enhancement is transferred into an adaptive intensity stretching process in the YUV domain. A different enhancement process is designed for a luminance (Y) channel and chrominance (UV) channel. The parameters in the enhancement process are estimated based on the contents of the image. After the stretching-based enhancement, an adaptive unsharp masking process is applied to the luminance data. The quality is significantly improved for the images shot under unfavorable conditions.

Abstract: System and methods for estimating a vanishing point within an image, including comprising: programming executable on a processor for computing line segment estimation of one or more lines in said image, wherein one or more of the lines comprise multiple line segments as a single least-mean-square-error (LMSE) fitted lines. Additionally the one or more lines having multiple line segments are represented as a single least-mean-square-error (LMSE) fitted line, and the one or more lines are intersected to locate a vanishing point in a density space.

Abstract: A method and apparatus is described here that categorizes images by extracting regions and describing the regions with a 16-dimensional subscene feature vector, which is a concatenation of color, texture, and spatial feature vectors. By comparing the spatial feature vectors in images with similarly-obtained feature vectors in a Gaussian mixture based model pool (obtained in a subscene modeling phase), the images may be categorized (in a subscene recognition phase) with probabilities relating to each region or subscene. Higher probabilities are likelier correlations. The device may be a single or multiple core CPU, or parallelized vector processor for characterizing many images. The images may be photographs, videos, or video stills, without restriction. When used real-time, the method may be used for visual searching or sorting.

Abstract: A method and apparatus for forming a demosaiced image from a color-filter-array (“CFA”) image is provided. The CFA image comprises a first set of pixels colored according to a first (e.g., a green) color channel, a second set of pixels colored according to a second (e.g., a red) color channel and a third set of pixels colored according to a third (e.g., blue) color channel. The method may include obtaining an orientation map, which includes, for each pixel of the color-filter-array image, an indicator of orientation of an edge bounding such pixel. The method may further include interpolating the first color channel at the second and third sets of pixels as a function of the orientation map so as to form a fourth set of pixels.

Abstract: A cathode ray tube (CRT) test system for testing a CRT device is disclosed. The CRT test system comprises a connecting device, a power module, a control module, a first test module, and an illumination module. The connecting device detachably and electrically connects with the CRT device; the power module electrically connects with the connecting device; the control module electrically connects with the connecting device; and the control module comprises test program. After the CRT device receives a control signal transmitted from the test program, the CRT device generates a response signal and transmits the response signal to the first test module. When the first test module detects that the control signal and the response signal fit in with a first test signal status, the illumination module displays a first light sign; when the control signal and the response signal do not fit in with the first test signal status, the illumination module displays a second light sign.

Abstract: Content adaptive detection of images having stand-out objects involves block variance-based detection and determining if an object includes a stand-out object. The images with a stand-out object are further processed to isolate an object of interest. The images without a detected stand-out object are further processed with a transition map-based detection method which includes generating a transition map. If an object portrait is determined from the transition map, then the image is further processed to isolate the object of interest.

Abstract: A method and system to improve the performance of phase correlation motion estimation for low-bit-precision implementation are described herein. Phase correlation uses the Fast Fourier Transform (FFT) with operations with infinite-precision constants. Since physical implementations use finite-precision arithmetic, there is some loss in precision relative to the ideal infinite-precision case. In low-complexity implementations, it is desirable to use as few bits as possible, and if the precision is too low, the performance of traditional phase correlation suffers. A pre-processing technique is applied to the data prior to taking the FFT, which minimizes the negative effects of finite precision in the FFT and allows high quality results from phase correlation. The pre-processing step is a content-dependent contrast adjustment that maps the range of the input images' pixel values to the range of input values for the FFT. There is no post-processing required after the FFT to compensate for the pre-processing step.

Abstract: Phase correlation is an established method for computing motion which relies on the ability to find peaks in a computed phase correlation surface. Two methods to improve the ability to detect peaks in the phase correlation surface are described herein. The first method applies a theoretically-derived and spatially-varying gain to the phase correlation surface. The gain compensates for peaks whose amplitudes have been decreased due to windowing effects; such effects are unavoidable in phase correlation. The second method uses concepts from matched filters to improve detection of peaks whose amplitudes are diminished due to a spreading of the peak energy into surrounding positions in the phase correlation surface. Peak detection filters allow such low-amplitude peaks to be properly detected. It is possible to use only the first method, or only the second method or both methods combined.

Abstract: A method of improving accuracy and reliability of motion estimation is described herein. In one aspect, a 2D neighborhood of phase correlation peak is approximated with an outer-product of two 1D vectors to eliminate the sub-pixel error. In another aspect, estimation of reliability is improved. In yet another aspect, two-pass phase correlation is implemented to eliminate sub-pel motion bias.

Abstract: A method of image color tone correction is disclosed, which may include the steps of (a) providing a reference image and one or more source images, wherein all images are comprised of pixels of a scene; (b) correcting a designated portion of one of the source images based on the reference image to create a color tone corrected image; and (c) outputting the color tone corrected image to a computer readable medium. The designated portion may be any subset of one of the source images, and may be a simple cropping, interlace, or other form of selection, such as areas of interest in the scene. Additionally, the designated portion may be used for determining a set of coefficients for overall color image correction of the complement of the source images. This method brings the increased tonal and saturation qualities of still images to those in motion.

Abstract: A method of and apparatus for generating a reference line or a virtual level enables tilt correction of an image. A user is able to utilize the reference line to see how an image should be oriented for the objects within the image to be level. The user is able to then correct the tilt of the image as desired using the reference line for assistance.

Abstract: A method of and an apparatus for automatically estimating the tilting angle in tilted images enables level correction of the images. A preferred orientation of objects in an image and the deviation of the current orientation from the preferred orientation is determined by tilt image analysis without object recognition. Tilt image analysis includes several steps such as gradient feature computation, line segment tracking, line segment estimation and orientation deviation estimation. Once the tilt angle is determined, the image can be corrected so that an object or scene is not tilted or is tilted by only the appropriate amount.

Abstract: A gesture recognition system and method that inputs videos of a moving hand and outputs the recognized gesture states for the input sequence. In each image, the hand area is segmented from the background and used to estimate parameters of all five fingers. The system further classifies the hand image as one of the postures in the pre-defined database and applies a geometric classification algorithm to recognize the gesture. The system combines a skin color model with motion information to achieve real-time hand segmentation performance, and considers each dynamic gesture as a multi-dimensional volume and uses a geometric algorithm to classify each volume.

Abstract: A gesture recognition system using a skin-color based method combined with motion information to achieve real-time segmentation. A Kalman filter is used to track the centroid of the hand. The palm center, palm bottom, as well as the largest distance from the palm center to the contour from extracted hand mask are computed. The computed distance to a threshold is then compared to decide if the current posture is “open” or “closed.” In a preferred embodiment, the transition between the “open” and “closed” posture to decide if the current gesture is in “select” or “grab” state.

Abstract: Reducing artifacts and improving quality for image and video encoding is performed in one pass to preserve natural edge smoothness and sharpness. To reduce artifacts and improve quality, several steps are implemented including spatial variation extraction, determining if a block is flat or texture/edge, classifying the pixels as texture or noise, detecting a dominant edge, checking the spatial variation of neighboring blocks, generating base weights, generating filter coefficients, filtering pixels and adaptive enhancement. A device which utilizes the method of reducing artifacts and improving quality achieves higher quality images and/or video with reduced artifacts.