Abstract: A subjective information record for inputting data that enables subjective information about a multimedia content to be represented with well defined syntax and semantics and linked to the multimedia content. The record includes a field for inputting data that identifies a multimedia content and a location of the multimedia content; a field for inputting data that identifies an author of subjective information about the multimedia content; a field for a language of the subjective information; and a field for subjective information about the content.

Abstract: A video conferencing system and method which automatically determines the appropriate preset camera parameters corresponding to participants participating in the video conference. A camera zooms out or pans the video conference space and looks for participants based on their faces. When a participant is detected, the preset camera parameters for that participant are calculated for when the center of the participant is in the center of the camera's view. This is continued for all the participants in the room. The optimal position for each participant and corresponding camera parameters are determined based on cultural preferences. Updates in the presets can be made periodically by the camera zooming out or panning the room. Multiple cameras can be used to continually update the presets.

Abstract: An image authentication system utilizes a coded version of a digital image signature and embeds that information into a transform representation of the image. A digital image signature preferably is extracted from the image data. That signature is coded so that the signature is not immediately recognizable. The coded information is embedded into high frequency coefficients in a transform representation of the image. The contents of the high frequency coefficients of a received or later accessed signal can then be deciphered and compared to the signature determined directly from the image to authenticate the integrity of the signal.

Abstract: A video retrieval system is presented that allows a user to quickly and easily select and receive stories of interest from a video stream. The video retrieval system classifies stories and delivers samples of selected stories that match each user's current preference. The user's preferences may include particular broadcast networks, persons, story topics, keywords, and the like. Key frames of each selected story are sequentially displayed; when the user views a frame of interest, the user selects the story that is associated with the key frame for more detailed viewing. This invention is particularly well suited for targeted news retrieval. In a preferred embodiment, news stories are stored, and the selection of a news story for detailed viewing based on the associated key frames effects a playback of the selected news story. The principles of this invention also allows a user to effect a directed search of other types of broadcasts as well.

Abstract: A method for placing overlay elements in images of a digital video so that desirable viewing areas of the images are not occluded by the overlay element, the method includes extracting frames from the video, each of the frames defining an image which may have one or more desirable viewing areas. For each frame, an area is selected in the image for placing an overlay element and the percentage the overlay element overlaps the desirable viewing areas is determined for the selected area. The overlay element is placed in the selected area if the percentage is below a maximum allowed predetermined percentage of overlapping. If it is not, one or more other areas in the image are selected and/or the size, shape, aspect ratio of the overlay element is changed until a position and/or size, shape, aspect ratio is found which minimizes occlusion of the desirable viewing areas in the image.

Abstract: In order to detect a face disposed within a digital image, the pixels of the image are grouped based on whether they are skin color. The edges of the skin colored areas are removed by eliminating pixels that have surrounding pixels with a high variance in the luminance component. The resulting connected components are classified to determine whether they could include a face. The classification includes examining: the area of the bounding box of the component, the aspect ratio, the ratio of detected skin to the area of the bounding box, the orientation of elongated objects, and the distance between the center of the bounding box and the center of mass of the component. Components which are still considered facial candidates are mapped on to a graph. The minimum spanning trees of the graphs are extracted and the corresponding components which remain are again classified for whether they could include a face.

Abstract: A method for placing overlay elements in images of a digital video so that desirable viewing areas of the images are not occluded by the overlay element, the method includes extracting frames from the video, each of the frames defining an image which may have one or more desirable viewing areas. For each frame, an area is selected in the image for placing an overlay element and the percentage the overlay element overlaps the desirable viewing areas is determined for the selected area. The overlay element is placed in the selected area if the percentage is below a maximum allowed predetermined percentage of overlapping. If it is not, one or more other areas in the image are selected and/or the size, shape, aspect ratio of the overlay element is changed until a position and/or size, shape, aspect ratio is found which minimizes occlusion of the desirable viewing areas in the image.

Abstract: A subjective information record for inputting data that enables subjective information about a multimedia content to be represented with well defined syntax and semantics and linked to the multimedia content. The record includes a field for inputting data that identifies a multimedia content and a location of the multimedia content; a field for inputting data that identifies an author of subjective information about the multimedia content; a field for a language of the subjective information; and a field for subjective information about the content.

Abstract: In order to detect a face disposed within a digital image, the pixels of the image are grouped based on whether they are skin color. The edges of the skin colored areas are removed by eliminating pixels that have surrounding pixels with a high variance in the luminance component. The resulting connected components are classified to determine whether they could include a face. The classification includes examining: the area of the bounding box of the component, the aspect ratio, the ratio of detected skin to the area of the bounding box, the orientation of elongated objects, and the distance between the center of the bounding box and the center of mass of the component. Components which are still considered facial candidates are mapped on to a graph. The minimum spanning trees of the graphs are extracted and the corresponding components which remain are again classified for whether they could include a face.

Abstract: A digital image is divided into a plurality of cells having a first sequence. A random seed is generated and used to produce two sets of pseudo-random numbers. The first set of pseudo-random numbers are used to alter the location and shape of the cells thereby creating a new set of cells that the image is divided into. A measurement is taken for each of these new cells. The second set of pseudo-random numbers creates a second sequence. Each of the new cells corresponding to the first sequence is compared to another new cell corresponding to the second sequence. This comparison is related to a threshold and yields a fingerprint. The resultant fingerprint is transmitted along with the image and the random seed. A receiver performs the same algorithm on its received image. If it produces the same fingerprint as the one it receives, it is assumed that the image has not been altered.

Abstract: In order to detect a face disposed within a digital image, the pixels of the image are grouped based on whether they are skin color. The edges of the skin colored areas are removed by eliminating pixels that have surrounding pixels with a high variance in the luminance component. The resulting connected components are classified to determine whether they could include a face. The classification includes examining: the area of the bounding box of the component, the aspect ratio, the ratio of detected skin to the area of the bounding box, the orientation of elongated objects, and the distance between the center of the bounding box and the center of mass of the component. Components which are still considered facial candidates are mapped on to a graph. The minimum spanning trees of the graphs are extracted and the corresponding components which remain are again classified for whether they could include a face.

Abstract: The invention is a method and apparatus for determining similarity between images based upon a count of the number of similar characterizations of the images. To provide for efficiency in the comparison process, indexed lists of image identifiers are maintained, and the count of similar characterizations of an image is determined by the count of occurrences of the image's identifier in selected lists. The selected lists are determined by a characterization of a target image from which similar images are to be identified. The indexing and retrieval techniques presented herein are suitable for a variety of image characterization techniques, such as characterizations based upon the color content or edge content of partitions of an image. Multiple indexes can be associated with one or more characteristic measures of each partition, allowing for image retrieval based on one or more characteristics of the target image.

Abstract: The system builds a database which stores data corresponding to a plurality of images. To begin, the system divides each image into N (N≧1) regions. Then, for each of the N regions, the system calculates a histogram of the region, generates a binary representation of the histogram, and stores data corresponding to the image in a binary tree based on the binary representation. The database may then be used to determine images which are similar to a query image. To do this, the system divides the query image into N regions, each of which corresponds to one of the binary trees in the database. Data corresponding to one or more images in the database is then retrieved from the binary trees based on the N regions. The system then determines which of these images is similar to the query image based on the retrieved data.

Abstract: An image retrieval system for retrieving images from a data base in dependence upon the degree of correspondence with a user-input target image. The target image and each image in the data base are characterized by a vector of edge characteristics within each image. Retrieval is effected by a comparison of the characteristic vectors, rather than a comparison of the images themselves. To optimize performance, a set of characteristic vectors, at different levels of abstraction of the image, are associated with each data base image; images which are similar to the target image, at any level of abstraction, are retrieved. To further optimize the speed of search, the data base is partitioned into clusters, and the search is limited to the cluster most similar to the target image.

Abstract: A digital radiologic image, notably a mammogram, is automatically processed by a computer to identify suspect masses. The identification is done by thresholding at least a region of interest of the image at, at least, 20 threshold levels determined from a histogram of the image to discriminate spots, and classifying the spots by size, shape and variance in intensity of the pixels comprising the spot. A processed mammogram having the suspect masses marked or enhanced is produced and displayed.

Abstract: Two digital radiologic images taken from different viewing directions of a same region, notably a mammogram study of the same breast, are automatically processed by a computer to identify suspect masses by producing processed mammogram images in which suspect spots, the skinline and the nipple are marked or enhanced. Candidate suspect spots are initially identified by individually processing each mammogram image separately. Then characteristics of the candidate spots identified in the mammogram images taken from different viewing directions are compared to eliminate false positives. The characteristics compared include position of the spots relative to the explicitly detected nipple, and their size, shape, brightness and brightness variance.

Abstract: Digital mammograms are automatically segmented into background, corresponding to the region external of the breast, and foreground, corresponding to the region within the breast, by the detection of the skinline which forms a border between these regions. A binary array is produced representing an initial coarse segmentation between foreground and background in which a binary one value is assigned to each pixel in the binary array whose intensity exceeds an intensity threshold and/or whose gradient amplitude exceeds a gradient threshold.