University of Pennsylvania

Shih-Schon Lin

Nader Engheta

Edward N Pugh, Jr

Konstantin Yemelyanov

A method for digital processing comprises receiving a first set of image data corresponding to a first time and a second set of image data corresponding to a different time. Intensity values for each of a first pixel and second pixel are extracted from each of the first and second set of image data. Substantially linear values in a plurality of intensity values are identified for each of the first and second pixels. For each of the first and second pixels, regression is performed on each of the substantially linear values in the plurality of intensity values. The regression may calculate an intensity level as a function of exposure time. The regression may generate slopes corresponding to the linear values for each of the first and second pixels. These slopes more accurately represent the actual ratio of intensity values. The ratios are useful in applications, such as, for example, polarization.

High linear dynamic range imaging

Shadow is an inseparable aspect of all natural scenes. When there are multiple light sources or multiple reflections several different shadows may overlap at the same location and create complicated patterns. Shadows are a potentially good source of information about a scene if the shadow regions can be properly identified and segmented. However, shadow region identification and segmentation is a difficult task and improperly identified shadows often interfere with machine vision tasks like object recognition and tracking. A shadow separation and contrast enhancement method based on the polarization of light is provided. Polarization information of scenes is captured by a polarization-sensitive camera and the scenes are processed to effectively separate shadows from different light sources.

Separation and contrast enhancement of overlapping cast shadow components and target detection in shadow using polarization

An optical fingerprinting method extracts high quality latent fingerprints from a surface without any invasive chemical or physical contact with the examined object, and requires no cooperation of the subject. Rather than employing extraneous material, the optical properties of the latent fingerprint are used to generate one or more images with sufficient contrast to distinguish the latent fingerprint or some other deformation in the surface. The system includes a light source oriented to apply light at an angle of incidence to the surface at the position to be examined for the latent fingerprint or deformation, a camera oriented to receive light specularly and diffusely reflected from the surface and/or by the fingerprint or deformation on the surface, and a processor that performs the computation for digital contrast enhancement and/or reprojection of the recovered fingerprint image to a frontal view if necessary. The technique uses optical polarization properties to enhance the images by placing a linearly polarized filter(s) in front of the observing camera. At least two pictures of the same scene with the same lighting and view angle arrangement are taken whereby each of the pictures differ only in that the orientations of the polarization filter are different. At least two light polarization parameters for each pixel are computed from the two or more images taken with different polarizer orientations. An image with each pixel value representing the value of one of the polarization parameters or a function of the polarization parameters is generated and displayed, with some digital contrast enhancement and/or reprojection applied. The hidden latent fingerprint pattern is revealed in at least one such image with the interfering background pattern significantly suppressed.

Growing community of inventors

Philadelphia, PA, United States of America

Shih-Schon Lin