Method for inverting motion blur of an image captured in a multiple camera system
The present invention provides a system and method for utilizing multiple camera systems including at least three cameras, each camera having at least partly overlapping fields for removing blur in a moving object relative to a static background captured by the multiple camera system, comprising separating and extracting the moving object from the static background by isolating pixels corresponding to the moving object and distinguishing them from the static background by using images captured by the at least three cameras having field of views covering the moving object, enhancing clarity of the separated moving object by deblurring the separated moving, blurring the static background, reintegrating the deblurred separated moving object onto the blurred static background in the position of the images where the separated moving object was extracted.
1 . A method for utilizing a multiple camera system including at least three cameras, each camera having at least partly overlapping fields for removing blur in a moving object relative to a static background captured by the multiple camera system, characterized in, separating and extracting the moving object from the static background by isolating pixels corresponding to the moving object and distinguishing the isolating pixels from the static background by using images captured by the at least three cameras having field of views covering the moving object, enhancing clarity of the separated moving object by deblurring the separated moving, blurring the static background, and reintegrating the deblurred separated moving object onto the blurred static background in the position of the images where the separated moving object was extracted.
2 . A method according to claim 1 , characterized in that the separated and extracted moving object includes pixels in a border area behind the moving object relative to the direction of the moving object which is a mixture of pixels from both the moving object and the static background.
3 . A method according to claim 2 , characterized in that the mixture of pixels from both the moving object and the static background are blended into the blurred static background when the deblurred separated moving object is reintegrated onto the blurred static background.
4 . A method according to claim 1 , characterized in that the step of deblurring includes reversing motion blur by using convolution and deconvolution of a Point Spread Function (PSF) derived from velocity amplitude and direction of the moving object together with its distance to the cameras capturing the moving object.
5 . A method according to claim 4 , characterized in that the distance to the moving object from the at least three cameras capturing the moving object is derived from the captured images.
6 . A method according to claim 1 , characterized in that the step of blurring the static background includes using a Gaussian blur function on the pixels of the static background.
7 . A multiple camera system comprising at least three cameras, each camera having at least partly overlapping fields of view for removing blur in a moving object relative to a static background, characterized in comprising a processor device adjusted to separate and extract the moving object from the static background by isolating pixels corresponding to the moving object and distinguishing the isolating pixels from the static background by using images captured by the at least three cameras having a field of view covering the moving object, enhance clarity of the separated moving object by deblur of the separated blurring the static background, and reintegrating the deblurred separated moving object onto the blurred static background in the position of the images where the separated moving object was extracted.