Abstract: A computer implemented method for fusing images taken by a plurality of cameras is disclosed, comprising the steps of: receiving a plurality of images of the same scene taken by the plurality of cameras; generating Laplacian pyramid images for each source image of the plurality of images; applying contrast normalization to the Laplacian pyramids images; performing pixel-level fusion on the Laplacian pyramid images based on a local salience measure that reduces aliasing artifacts to produce one salience-selected Laplacian pyramid image for each pyramid level; and combining the salience-selected Laplacian pyramid images into a fused image. Applying contrast normalization further comprises, for each Laplacian image at a given level: obtaining an energy image from the Laplacian image; determining a gain factor that is based on at least the energy image and a target contrast; and multiplying the Laplacian image by a gain factor to produce a normalized Laplacian image.

Abstract: The present invention provides a method and a system for high performance image signal processing of continuous images in real time. The system comprising a focal plane array for generating continuous source image frames in real time. The focal plane array divided logically into blocks of sub-frames. The system also comprising an analog to digital converter (ADC) layer having an array of ADC elements for converting the source image frames into a digital data. The system further comprising a digital processor layer having an array of processing elements for processing the digital data and an interconnecting layer for connecting each one of the ADC elements and the digital processing elements substantially vertically to the focal plane and substantially parallel to one another. The processing comprising reducing image motion blur, increasing image dynamic range, increasing image depth of field and obtaining features of the images.

Abstract: A method and apparatus for optimizing image quality based on scene content comprising a sensor for generating a sequence of frames where each frame in the sequence of frames comprises content representing a scene and a digital processor, coupled to the sensor, for performing scene content analysis and for establishing a window defining a number of input frames from the sensor and processed output frames, and for aligning and combining the number of frames in the window to form an output frame, wherein sensor parameters and frame combination parameters are adjusted based on scene content.

Abstract: A computer implemented method for fusing images taken by a plurality of cameras is disclosed, comprising the steps of: receiving a plurality of images of the same scene taken by the plurality of cameras; generating Laplacian pyramid images for each source image of the plurality of images; applying contrast normalization to the Laplacian pyramids images; performing pixel-level fusion on the Laplacian pyramid images based on a local salience measure that reduces aliasing artifacts to produce one salience-selected Laplacian pyramid image for each pyramid level; and combining the salience-selected Laplacian pyramid images into a fused image. Applying contrast normalization further comprises, for each Laplacian image at a given level: obtaining an energy image from the Laplacian image; determining a gain factor that is based on at least the energy image and a target contrast; and multiplying the Laplacian image by a gain factor to produce a normalized Laplacian image.

Abstract: A computer implemented method for fusing images taken by a plurality of cameras is disclosed, comprising the steps of: receiving a plurality of images of the same scene taken by the plurality of cameras; generating Laplacian pyramid images for each source image of the plurality of images; applying contrast normalization to the Laplacian pyramids images; performing pixel-level fusion on the Laplacian pyramid images based on a local salience measure that reduces aliasing artifacts to produce one salience-selected Laplacian pyramid image for each pyramid level; and combining the salience-selected Laplacian pyramid images into a fused image. Applying contrast normalization further comprises, for each Laplacian image at a given level: obtaining an energy image from the Laplacian image; determining a gain factor that is based on at least the energy image and a target contrast; and multiplying the Laplacian image by a gain factor to produce a normalized Laplacian image.

Abstract: The present invention provides a method and a system for high performance image signal processing of continuous images in real time. The system comprising a focal plane array for generating continuous source image frames in real time. The focal plane array divided logically into blocks of sub-frames. The system also comprising an analog to digital converter (ADC) layer having an array of ADC elements for converting the source image frames into a digital data. The system further comprising a digital processor layer having an array of processing elements for processing the digital data and an interconnecting layer for connecting each one of the ADC elements and the digital processing elements substantially vertically to the focal plane and substantially parallel to one another. The processing comprising reducing image motion blur, increasing image dynamic range, increasing image depth of field and obtaining features of the images.

Abstract: Apparatus and a concomitant method of identifying the direction of motion within a scene that is represented by a sequence of images, e.g., a series of frames in a video sequence. The apparatus contains a flow field generator, a flow field segmentor, and alarm detector and an alarm processor. The flow field generator processes the sequence of images and generates a substantially distortionless flow field representing the motion within a scene. The flow field is a vector representation of the motion of the scene that represents both the magnitude and the direction of the motion. The flow field is generated by correlating at least two frames in the sequence of images. This flow field is analyzed by a flow field segmentor to determine the magnitude and direction of motion with the scene and segment the motion information from the static portions of the scene. An alarm detector then processes the motion information to determine if an alarm should be generated based upon the motion information.

Abstract: Apparatus and a concomitant method of identifying the direction of motion within a scene that is represented by a sequence of images, e.g., a series of frames in a video sequence. The apparatus contains a flow field generator, a flow field segmentor, and alarm detector and an alarm processor. The flow field generator processes the sequence of images and generates a substantially distortionless flow field representing the motion within a scene. The flow field is a vector representation of the motion of the scene that represents both the magnitude and the direction of the motion. The flow field is generated by correlating at least two frames in the sequence of images. This flow field is analyzed by a flow field segmentor to determine the magnitude and direction of motion with the scene and segment the motion information from the static portions of the scene. An alarm detector then processes the motion information to determine if an alarm should be generated based upon the motion information.

Abstract: A digital image warper system produces a warped output image from an input image. The warper system increases the sampling rate of a sampled image above the Nyquist rate or samples an analog image at a higher rate than the Nyquist rate and prefilters the upsampled image to counteract filtering which may occur during the warping operation. The upsampled image is warped using an interpolator for resampling to produce a warped upsampled image. The interpolator is, for example, a low quality interpolator such as a bilinear interpolator. The warped image is then down-sampled to the same resolution as the input image to produce the warped image. Down-sampling and warping can be combined into one step by modifying the geometric transformation function implemented when warping the upsampled image.

Abstract: Disclosed is a technique for deriving a composite video image by merging foreground and background video image data supplied from a plurality of separate video signal sources employing pattern-key insertion, rather than prior-art color-key insertion, for this purpose. Pattern-key insertion involves replacing a first pattern in a video scene with a second pattern. This is accomplished by first detecting the first pattern in the video scene and estimating the pose of this detected first pattern with respect to a reference object in the video scene. The second pattern is then geometrically transformed using the pose estimate of the detected first pattern. Finally, the detected first pattern is replaced with the geometrically-transformed second pattern.