Patents by Inventor Cassandra Turner Swain

Cassandra Turner Swain has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Patent number: 6393142
    Abstract: A method and apparatus provide a technique for generating a mesh based on an image view. The mesh includes a plurality of trapezoidal patches which can be used in a comparison operation with regard to a second co-planar image view. The comparison operation extracts disparity information between the patches of the respective images and these disparities can be used to calculate depth estimations for objects within the image of use.
    Type: Grant
    Filed: April 22, 1998
    Date of Patent: May 21, 2002
    Assignee: AT&T Corp.
    Inventors: Cassandra Turner Swain, Yao Wang
  • Patent number: 6381345
    Abstract: A simple method for segmenting eyes and extracting parameters enables further processing of the image to enable a person to appear to be making eye contact with another via a video conferencing system. This method is a first step for eye synthesis and gaze detection because it can automatically extract select eye parameters useful to these processes. Its advantage is that no a priori information is necessary to segment eyes, unlike modeling and neural network methods. The method of the present invention first blurs the image to make it easier to determine the location of the two eye regions in the image. The eyebrows are then eliminated based on the located eye regions. The eyes are then segmented and the eye parameters are extracted from the resulting image According to the present invention, the process applies a Gaussian filter, h(x, y), where g(x, y) is the resulting image and ƒ(x, y) is the original image.
    Type: Grant
    Filed: June 3, 1997
    Date of Patent: April 30, 2002
    Assignee: AT&T Corp.
    Inventor: Cassandra Turner Swain
  • Patent number: 6301385
    Abstract: To segment moving foreground from background, where the moving foreground is of most interest to the viewer, this method uses three detection algorithms as the input to a neural network. The multiple cues used are focus, intensity, and motion. The neural network consists of a two-layered neural network. Focus and motion measurements are taken from high frequency data, edges; whereas, intensity measurements are taken from low frequency data, object interiors. Combined, these measurements are used to segment a complete object. Results indicate that moving foreground can be segmented from stationary foreground and moving or stationary background. The neural network segments the entire object, both interior and exterior, in this integrated approach. Results also demonstrate that combining cues allows flexibility in both type and complexity of scenes. Integration of cues improves accuracy in segmenting complex scenes containing both moving foreground and background.
    Type: Grant
    Filed: December 17, 1998
    Date of Patent: October 9, 2001
    Assignee: AT&T Corp.
    Inventors: Tsuhan Chen, Cassandra Turner Swain
  • Patent number: 6157733
    Abstract: One or more monocular cues are extracted from an original image and combined to enhance depth effect. An original image is acquired and segmented into one or more objects. The objects are identified as being either in the foreground or the background, and an object of interest is identified. One or more depth cues are then extracted from the original image, including shading, brightness, blur and occlusion. The depth cues may be in the form of one or more intermediate images having an improved depth effect. The depth cues are then combined or applied to create an image with enhanced depth effect.
    Type: Grant
    Filed: April 18, 1997
    Date of Patent: December 5, 2000
    Assignee: AT&T Corp.
    Inventor: Cassandra Turner Swain
  • Patent number: 6035060
    Abstract: A method and apparatus for generating region frames from video frames are disclosed which employs an industry standard encoder to lessen the negative impact on the quality of the transmitted video sequence while consuming fewer bits. The invention utilizes image segmentation and color replacement techniques to create the region frames. Each region frame includes a subject region, zero or more previously segmented regions and zero or more non-subject regions. The subject region is defined by the pixels of the original video frame. The previously segmented regions and non-subject regions are assigned replacement pixels P.sub.n,y and C.sub.n, respectively. The replacement pixel C.sub.n is chosen to indicate a color that is not likely to be confused with any color in the subject region R.sub.n. The replacement pixels P.sub.n,y are chosen such that the compression ratio of the region frame data is maximized.
    Type: Grant
    Filed: February 14, 1997
    Date of Patent: March 7, 2000
    Assignee: AT&T Corp
    Inventors: Tsuhan Chen, Barin Geoffry Haskell, Cassandra Turner Swain
  • Patent number: 5960111
    Abstract: To segment moving foreground from background, where the moving foreground is of most interest to the viewer, this method uses three detection algorithms as the input to a neural network. The multiple cues used are focus, intensity, and motion. The neural network consists of a two-layered neural network. Focus and motion measurements are taken from high frequency data, edges; whereas, intensity measurements are taken from low frequency data, object interiors. Combined, these measurements are used to segment a complete object. Results indicate that moving foreground can be segmented from stationary foreground and moving or stationary background. The neural network segments the entire object, both interior and exterior, in this integrated approach. Results also demonstrate that combining cues allows flexibility in both type and complexity of scenes. Integration of cues improves accuracy in segmenting complex scenes containing both moving foreground and background.
    Type: Grant
    Filed: February 10, 1997
    Date of Patent: September 28, 1999
    Assignee: AT&T Corp
    Inventors: Tsuhan Chen, Cassandra Turner Swain
  • Patent number: 5890120
    Abstract: A synthesized signing image is generated which includes an original image of a subject and a sign image superimposed on the original subject image. The sign image is selected from a sign language database based on the speech (i.e., a letter or word) spoken by the subject. Based on the size and tone (intensity) of the subject's face in the original image, the database sign image is modified to match the intensity and have a proportional size to the subject's face. The modified database sign image is then superimposed on the original image of the subject to create a synthesized image in which the subject appears to be signing.
    Type: Grant
    Filed: May 20, 1997
    Date of Patent: March 30, 1999
    Assignee: AT&T Corp
    Inventors: Barin Geoffry Haskell, Cassandra Turner Swain
  • Patent number: 5710829
    Abstract: Motion video is represented by digital signals. The digital signals can be compressed by coding to reduce bitrate and thus save time and expense in transmitting and reproducing the video. The present invention is a system and method for reducing the amount of information required to satisfactorily reproduce video signals by distinguishing more pertinent portions of the video from less pertinent portions. In particular, edges are detected based on amplitude variations in the luminance characteristic of the picture elements. Sharp edges are deemed to be "focused". The portion of the video frame between the focused edges is deemed to be focused, as well. A template is created corresponding to the focused portion. A signal corresponding to the outline of the template is combined with the original frame signal to create a segmented frame signal. When motion detection information is available, a motion-based template may be created and intersected with the focus template.
    Type: Grant
    Filed: April 27, 1995
    Date of Patent: January 20, 1998
    Assignee: Lucent Technologies Inc.
    Inventors: Tsuhan Chen, Cassandra Turner Swain