Abstract: A method and system for automatically summarizing a video document. The video document is decomposed into scenes, shots and frames, and an importance value is assigned to each scene, shot and frame. Keyframes are allocated among the shots based on the importance value of each shot. The allocated number of keyframes are then selected from each shot. The number of keyframes may be altered for greater or lesser detail in response to user input.

Abstract: A system and method for merging scenes in a video sequence and constructing a keyframe to represent the underlying merged video content includes decomposing a video sequence into a series of component scenes, merging component scene pairs until a predetermined number of scene sets remain, extracting a keyframe from each scene set containing a single component scene, and constructing a new keyframe for each scene set containing a plurality of component scenes.

Abstract: A method of visualizing and retrieving data files includes displaying a plurality of images representing data files on a display device using a first distance metric between each data file. A portion of the images is redisplayed on the display device using a refined distance metric. Optionally, the method repeats the redisplaying step until a desired data file is identifiable. Then, the desired data file is retrieved.

Abstract: An apparatus and method for dither matrix design for color halftoning using dispersed dot clusters where the dot cluster centers are arranged in a homogeneous and isotropic pattern, in order avoid the moiré effect. The method requires that the size of the array and the number of dot seeds that will be in the array be predetermined.

Abstract: A method and system for automatically removing noise from images. First, a current pixel is received. Next, an activity metric of the current pixel is computed by using the current pixel and a neighborhood of pixels related to the current pixel. Then, a distance metric that indicates the likelihood that the current pixel is a background pixel is computed by using the activity metric and at least one background parameter. The distance metric provides a soft thresholding framework for the background removal. The current pixel is modified based on the distance metric.

Abstract: A method of operating a computer to produce contrast enhanced digital images commences with the step of producing a histogram of having a first axis corresponding to a measurable property (e.g., luminance) and a second axis corresponding to a count of pixels having a particular value for the measurable property. This histogram is divided into clusters and histogram equalization or stretching is performed on each cluster thereby producing a modified histogram. Using said modified histogram to adjust the value of said first measurable property in said digital form, thereby producing a contrast enhanced image. The histogram is divided into clusters using a pattern matching technique. For example, patterns in the histogram that resemble gaussian distributions and patterns that resemble uniform distributions are separated into individual clusters.

Abstract: A method and system for categorizing non-textual subject data, such as digital images, content-based data and meta-data to determine outcomes of classification tasks. The meta-data is indicative of the operational conditions of a recording device during the capturing of the content-based data. For example, the non-textual subject data may be a digital image captured by a digital camera, and the meta-data may include automatic gain setting, film speed, shutter speed, aperture/lens index, focusing distance, date and time, and flash/no flash operation. The subject image is tagged with selected classifiers by subjecting the image to a series of classification tasks utilizing both content-based data and meta data to determine classifiers associated with the subject image.

Abstract: A system and method is described for automatically determining in a scanned document image the presence of unwanted extraneous information caused by an extraneous device and scanner background information. Once the presence of this information is determined, the system and method of the present invention can compute, for instance, skew and crop statistics. From this, the image can be automatically deskewed and cropped appropriately without the background and extraneous information. The system and method accomplishes this by first determining the presence of unwanted extraneous and background information and then appropriately processing the document image. The extraneous information is ignored during deskew and crop computations. Also, the scanner background and the extraneous information are prevented from being included in the final digital representation of the image.

Abstract: An image compression system for compound images containing both text and pictures. The system is capable of receiving the images on a non-overlapping 8 by 8 pixel blank basis and includes a discrete cosine transformer connected to a quantizer drawing lossy quantization factors from quantization tables. The lossy quantization factors are modified by a variable quantization subsystem based on the frequency of changes in the block to provide low lossy quantization factors for high frequency of changes and high lossy quantization factors for low frequency of changes. The high frequency of changes being indicative of text and the low frequency of changes being indicative of pictures. The quantizer is connected to an entropy coder using lossless entropy encoding factors from Huffman tables to provide JPEG compliant files.

Abstract: A process and apparatus is described to improve the fidelity of compressed images by computing a scaling value for each block based on a perceptual classification performed in the spatial domain. This provides a computationally simple way to reduce artifacts by computing appropriate block-variable scale factors for the quantization tables used in frequency domain-based compression schemes such as the the JPEG compression standard. Because a scale factor for a block is determined based on computations performed in the spatial domain, such computations can be made in parallel with the Discrete Cosine Transform (DCT) computation, thereby providing the same throughput in hardware or parallel processing software as can be obtained by baseline JPEG.

Abstract: An image compression system for compound images containing both text and pictures. The system is capable of receiving the images on a non-overlapping 8 by 8 pixel blank basis and includes a discrete cosine transformer connected to a quantizer drawing lossy quantization factors from quantization tables. The lossy quantization factors are modified by a variable quantization subsystem based on the frequency of changes in the block to provide low lossy quantization factors for high frequency of changes and high lossy quantization factors for low frequency of changes. The high frequency of changes being indicative of text and the low frequency of changes being indicative of pictures. The quantizer is connected to an entropy coder using lossless entropy encoding factors from Huffman tables to provide JPEG compliant files.

Abstract: A system and method is described for automatically determining in a scanned document image the presence of unwanted extraneous information caused by an extraneous device and scanner background information. Once the presence of this information is determined, the system and method of the present invention can compute, for instance, skew and crop statistics. From this, the image can be automatically deskewed and cropped appropriately without the background and extraneous information. The system and method accomplishes this by first determining the presence of unwanted extraneous and background information and then appropriately processing the document image. The extraneous information is ignored during deskew and crop computations. Also, the scanner background and the extraneous information are prevented from being included in the final digital representation of the image.

Abstract: A system and method is described for automatically determining in a scanned document image the presence of unwanted extraneous information caused by an extraneous device and scanner background information. Once the presence of this information is determined, the system and method of the present invention can compute, for instance, skew and crop statistics. From this, the image can be automatically deskewed and cropped appropriately without the background and extraneous information. The system and method accomplishes this by first determining the presence of unwanted extraneous and background information and then appropriately processing the document image. The extraneous information is ignored during deskew and crop computations. Also, the scanner background and the extraneous information are prevented from being included in the final digital representation of the image.

Abstract: A process and apparatus is described to improve the digital processing of scanned negative films by reducing the amount of time necessary to perform the process and by increasing the robustness and quality of the images produced. These benefits are achieved by a process of color inversion, white point and black point mapping, and midtone adjustment. White and black mapping increases the dynamic range of the image, as well as removes the color cast of the negative film. A backlit image postprocessing algorithm can be employed which uses heuristics to identify backlighted situations, which are then brightened using a nonlinear power mapping. A midtone adjustment can include the sub-steps of contrast reduction and color adjustment. Contrast reduction reverses the film exposure characteristics. Color adjustment removes the remaining color cast in the midtone region of the image, and obtains the correct brightness.

Abstract: A red-eye reduction system is described that includes a masking module. The masking module converts an image into a mask having first state areas representing red color pixels of the image and second state areas representing other color pixels of the image. The image includes an eye with a red pupil. A pupil locating module is coupled to the masking module to locate a substantially first state area in the mask that resembles a pupil. A color replacing module is then coupled to the pupil locating module to change the red color pixels in the area into monochrome (grey) or other predefined colors. The color replacing module also adjusts the boundary of the area by changing the colors of pixels in close proximity to the area if the color of these pixels is determined to be sufficiently close to red such that natural appearance of the eye is maintained when reducing the red pupil. A method of reducing red-eye effect in a digital image is also described.

Abstract: An arrangement is described for determining the skew angle of a document image without requiring text or other skew detection information in the document or without human intervention. The arrangement accomplishes this by determining an edge of the document image within a scan image and uses that edge to determine the skew angle of the document image. The edge can be determined by locating the first or last document image pixel of each scan line of pixels in the scan image that belongs to the document image. This is accomplished by comparing each pixel of each scan line of pixels with a predetermined scan line of background pixels. The skew angle of the document image is then determined by computing the slope of the detected edge in the scan image. An arrangement of determining the boundary of the document image is also described.

Abstract: An image enhancing apparatus is described that includes a filter. The filter enhances the sharpness of an original image in accordance with a sharpening parameter. A sharpening parameter selection system is coupled to the filter. The sharpening parameter selection system determines the sharpening parameter for the filter such that the frequency distribution of a sharpened image of the original image by the filter is similar to that of a low resolution image of the original image. In addition, a method of sharpening an original image is also described.

Abstract: An image processing arrangement is described that minimizes moire patterns in scanned halftone images of a scanned image while maintaining the resolution of the scanned image. The arrangement includes a filtering processor that has a modified median filter with a specially shaped (e.g., cross-shaped) filter window. The filter window allows the modified median filter to preserve scanned text images of the scanned image when processing the scanned halftone images to minimize the moire patterns. In addition, the filtering processor determines if an image pixel of the scanned image appears to belong to scanned continuous tone images of the scanned image prior to processing the image pixel. The filtering processor does not process that image pixel and the pixel value of that image pixel is left unchanged if the image pixel appears to belong to the scanned continuous tone images.