USA as Represented by Secretary of the Navy

Kristofor B Gibson

Gregory Knowles Fleizach

A system and method for haze removal. The method comprises the steps of receiving, at a memory, an input image having pixels. The method further comprises converting, by a processor, each pixel in each channel of the input image to floating-point values in the range of zero to one [0,1], and brightness-correcting, by the processor, the input image to prevent the de-hazed output image from becoming overly dark. The method also includes estimating, by the processor, the airlight for the brightness-corrected image, calculating a transmission map for each color or intensity channel of the image, and refining the transmission map for each said color or intensity. A reduced-haze image is thereby provided.

Single image haze removal

A method involves generating a denoised image by performing a denoising operation on an input image of an object, the input image having turbulence and a contrast reducing scattering medium therein, generating a contrast enhanced image from the denoised image using a contrast enhancement algorithm, estimating a global motion vector of the contrast enhanced image using an image alignment algorithm, generating a contrast enhanced aligned image by globally aligning the contrast enhanced image to the global motion vector and locally aligning the contrast enhanced image with an optical flow method, temporally averaging the contrast enhanced aligned image, and generating an output image of the object by performing a deblur operation on the temporally averaged contrast enhanced aligned image. The method may further include measuring turbulence using sequences of input images and output images.

Joint contrast enhancement and turbulence mitigation method

A method involves using an adaptive wiener filter to generate a veiling estimate for a dark channel prior of each pixel in a single image having a contrast reducing scattering medium therein, where a global noise variance for the image is used to generate the veiling estimate and the global noise variance is naively estimated using all of the pixels in the image. A first contrast enhanced image is generated using a transmission calculated from the veiling estimate. The global noise variance is updated using statistics determined from the first contrast enhanced image, along with a second veiling estimate. A second contrast enhanced image is generated using the updated global noise variance, the second veiling estimate, and the local statistics from the original image.

Growing community of inventors

San Diego, CA, United States of America

Kristofor B Gibson