Image data coding and compression system utilizing controlled blurring

• US 5,453,844 A
• Filed: 11/17/1993
• Issued: 09/26/1995
• Est. Priority Date: 07/21/1993
• Status: Expired due to Term

• Current Assignee: University of Rochester
• Original Assignee: University of Rochester
• Inventors: Stossel, Bryan J., George, Nicholas
• Primary Examiner(s): Rogers, Scott A.
• Application Number:
  US08/154,311
• Time in Patent Office: 678 Days
• Field of Search: 358/426, 358/432, 358/433, 358/430, 358/428, 382/56, 380/18, 380/46, 380/44, 380/43
• US Class Current: 382/264
• CPC Class Codes: H04N 1/4486 using digital data encryption
• CPC Class Codes: H04N 19/60 using transform coding

Abstract

Data representing the image of an original picture having high resolution are first acquired. The data are then processed in accordance with a blurring function, which can be a randomly distributed point spread function. Processing can be carried out either in object space or in spatial-frequency space. In object space a multi-dimensional convolution is required. In the spatial frequency domain, processing is performed by a pointwise product of data representing the Fourier transform of the original image and the Fourier transform of the blurring function. The resulting blurred image is thereby coded such that the original scene is no longer discernible, but can be recovered provided the blurring function is known. Additionally, this blurred image can be shown to contain the detail of the original but can be stored or transmitted as data which are compressed by compression ratios which can range from 2 to 100. The compressed data can be restored to provide an image with detail like the original by inverse filtering using data representing the Fourier transform of the blurring function.

Claims

1. A system for compressing and decompressing image data signals representing an original image characterized by p(x,y), where x and y are the spatial coordinates of pixels of the image, which system comprises:

(a) means to convolve said image data signals with coded random pattern data signals generated according to a blurring function ##EQU11## where N is an integer greater than one and x_n and y_n are the positions of delta-function impulses, x_n and y_n being chosen from distributions characterized by probability density functions, to provide convolved data signals representing a blurred image;

(b) means to compress said convolved data signals to permit efficient storage or transmission of the information in said original image; and

(c) means to recover said original image from said convolved compressed data signals with a function which is the inverse of said compressing means and said coded blurring means to provide data signals representing said original image in restored form.

2. The system according to claim 1 wherein said signal convolving means, said compressing means, and said recovering means are operative in object space, said convolving means providing the convolution, g_b (x,y)=p(x,y)*b(x,y).

3. The system according to claim 1 wherein said signal convolving means, said compressing means, and said recovering means are operative in spatial frequency space.

4. The system according to claim 3 wherein

(a) said signal convolving means and compressing means comprises means for providing the Fourier transform of the data signals representing p(x,y) and b(x,y) which in said spatial frequency space are p(f_x,f_y) and B(f_x,f_y), respectively;

means for multiplying P(f_x,f_y) and B(f_x,f_y) to provide the product thereof, G_b (f_x,f_y); and

means for providing said compressed convolved data signals in response to and representing G_b (f_x,f_y); and

(b) said recovering means includes means for providing, in response to the compressed data signals representing G_b (f_x,f_y) and data signals representing B(f_x,f_y), data representing the quotient thereof, and means for providing the inverse Fourier transform of said quotient to provide said data signals representing said original image in restored form.

5. The system according to claim 1 further comprising means operative according to a random process for obtaining x_n and y_n.

6. The system according to claim 5 wherein said random process operative means is a generator providing a generally random point spread function of N points in an area in x and y containing said image.

7. The system according to claim 6 wherein N is in a range from 2 to 250.

8. The system according to claim 1 wherein said means to compress said convolved data signals further comprises either means for subsampling said signals after convolving with the blurring function or means for reducing the number of bits representing pixels at each of said spatial coordinates x &

y.

9. The system according to claim 8 wherein said data compressing means operative on pixels of said blurred image is selected from the group consisting of means for subsampling said pixels and means for reducing the number of bits of data signals representing said blurred pixels.

10. The system according to claim 1 wherein the blurring function is another blurring function which is described by ##EQU12## where s(x,y) is a sampling function including Gaussian and triangle functions.

11. The system according to claim 10 wherein s(x,y) is ##EQU13## in which w is a constant.

12. The system according to claim 10 wherein one of the impulses is taken at the origin i.e. δ

(x)δ

(y) and includes the point (x₁ =0,y₁ =0).

13. The system according to claim 10 wherein an impulse function at the origin Cδ

(x)δ

(y) is added to the generalized controlled blurring function b₁ (x,y) to give b₂ (x,y) defined by;

space="preserve" listing-type="equation">b.sub.2 (x,y)=Cδ

(x)δ

(y)+b.sub.1 (x,y).

14. The system according to claim 13 wherein the blurring function is:

##EQU14##

15. The system according to claim 1 wherein an impulse at the origin Cδ

(x)δ

(y) is added to define the blurring function.

16. The system according to claim 1 further comprising means to store or to transmit said compressed convolved data signals.

25. The system according to claim 9 further comprising means to store or to transmit said second array of pixels.

17. A system for representing an original image having a resolution characterized by a first array of picture elements (pixels) with a second array of pixels less in number than the pixels in said first array and from which said second array a restored version of said original image can be reproduced with approximately the same detail as said original image, said system comprising means for spatially coding said original image according to a spatial code to provide a blurred image, and means for spatially decoding said blurred image responsive to said spatial code to provide said restored version of said original image.

18. The system according to claim 17 further comprising means for translating said blurred image into said second array, and said spatial decoding means including means responsive to said spatial code which codes said image for translating said second array into an array of pixels generally corresponding to said first array.

19. The system according to claim 18 wherein said spatial code is a generally random point spread function of N points where N is an integer greater than one.

20. The system according to claim 19 wherein N is in a range from 2 to 250.

21. The system according to claim 18 wherein said translating means comprises means for scanning said blurred image to provide a pattern of pixels forming said second array.

22. The system according to claim 18 wherein said translating means comprises means for sampling said blurred image to provide data signals representing said second array.

23. The system according to claim 17 wherein said spatial coding means includes means operative upon the Fourier transform of said original image in accordance with a selected speckle pattern to provide said blurred image, said spatial decoding means includes means operative on said blurred image in accordance with said selected speckle pattern.

24. The system according to claim 23 wherein said speckle pattern is represented by the Fourier transform of a plurality of randomly distributed points, N in number where N is an integer greater than one.

26. A method for representing an original image having a resolution characterized by a first array of picture elements (pixels) with a second array of pixels less in number than the pixels in said first array and from which said second array a restored version of said original image can be reproduced with approximately the same detail as said original image comprising the steps of:

spatially coding said original image according to a spatial code to provide a blurred image, and

spatially decoding said blurred image in response to said spatial code to provide said restored version of said original image.

27. A method according to claim 26 further comprising translating said blurred image into said second array, and where said spatially decoding step includes translating said second array into an array of pixels generally corresponding to said first array.

28. A method according to claim 26 wherein said spatial code is a generally random point spread function of N points where N is an integer greater than one.

29. A method according to claim 26 wherein said step of spatially coding according to said spatial code includes operating upon the Fourier transform of said original image in accordance with a selected speckle pattern to provide said blurred image, and said step of spatially decoding in response to said spatial code includes operating on said blurred image in accordance with said selected speckle pattern.

30. A method according to claim 26 whereby said spatial code is a blurring function represented by ##EQU15## where x and y are spatial coordinates of pixels, N is an integer greater than one and x_n and y_n are the positions of delta-function impulses, x_n and y_n being chosen from distributions characterised by probablity density functions, to provide convolved data signals representing a blurred image.

31. A method according to claim 30 wherein said blurring function is described by ##EQU16## where s(x,y) is a sampling function including Gaussian and triangle functions.