Method, data processing system, computer program product and computer readable medium for determining image sharpness
The invention is a method for determining whether an input image is sufficiently sharp, comprising the steps of providing a blur score threshold value, inputting the input image to an image processing system, generating, by the image processing system, a two-dimensional frequency spectrum of the input image, generating a one-dimensional frequency spectrum from the two-dimensional frequency spectrum, fitting a straight line on the one-dimensional frequency spectrum, and determining a blur score value based on a residual of the fitting, and considering the input image as sufficiently sharp based on a comparison of the blur score value with the blur score threshold value. The invention further relates to a data processing system, a computer program product and a computer readable medium carrying out the above method.
1 . A method for determining whether an input image is sufficiently sharp, comprising the steps of
providing a blur score threshold value,
inputting the input image to an image processing system,
generating, by the image processing system, a two-dimensional frequency spectrum of the input image,
generating a one-dimensional frequency spectrum from the two-dimensional frequency spectrum,
fitting a straight line on the one-dimensional frequency spectrum, and
determining a blur score value based on a residual of the fitting, and considering the input image as sufficiently sharp based on a comparison of the blur score value with the blur score threshold value,
wherein providing the blur score threshold value comprises making a decision on the basis of empirical data between a first scenario and a second scenario, wherein in said first scenario, the blur score threshold value is a predetermined value, and in said second scenario, the blur score threshold value is determined by the following steps:
generating a blurred image from the input image,
generating, by an image processing system, a further two-dimensional frequency spectrum of the blurred image,
generating a further one-dimensional frequency spectrum from the further two-dimensional frequency spectrum, and
fitting a further straight line on the further one-dimensional frequency spectrum, wherein the blur score threshold value is based on a residual of the fitting.
2 . The method according to claim 1 , characterized in that the one-dimensional frequency spectrum is generated from a two-dimensional amplitude spectrum of the two-dimensional frequency spectrum.
3 . The method according to claim 2 , characterized by calculating a logarithm of the two-dimensional amplitude spectrum, wherein the one-dimensional frequency spectrum is generated from the logarithm of the two-dimensional amplitude spectrum.
4 . The method according to claim 2 , characterized in that the one-dimensional frequency spectrum is a histogram having bins corresponding to frequency ranges of the two-dimensional frequency spectrum and histogram values for each bin, wherein each histogram value is an average amplitude or an integrated amplitude within the corresponding frequency range.
5 . The method according to claim 1 , characterized in that the two-dimensional frequency spectrum is centralized, and the one-dimensional frequency spectrum is generated by radial averaging.
6 . The method according to claim 1 , characterized in that the two-dimensional frequency spectrum of the input image is generated by a two-dimensional Fourier-transform or a two-dimensional wavelet transform.
7 . The method according to claim 1 , characterized in that the straight line is fitted by a linear regression method.
8 . The method according to claim 1 , characterized in that in the second scenario the blurred image is generated by a blurring having a blur parameter, and the blur parameter is determined by the following steps:
taking a sharp image,
generating a series of blurred images based on the sharp image by applying a series of subsequent blurring,
calculating a blur score value for each one of the series of blurred images, and
selecting the blur parameter corresponding to a minimum of the blur score values of the series of the blurred images.
9 . The method according to claim 1 , characterized in that the blurred image is generated by a Gaussian blurring.
10 . The method according to claim 8 , characterized in that the further two-dimensional frequency spectrum is generated the same way as the two-dimensional frequency spectrum, the further one-dimensional frequency spectrum is generated the same way as the one-dimensional frequency spectrum, and the further straight line is fitted the same way as the straight line.
11 . The method according to claim 1 , characterized in that the comparison of the blur score value and the blur score threshold value is implemented by generating a difference score value by subtracting the blur score threshold value from the blur score value, wherein difference score values having a positive value or a non-negative value indicate that the input image is sufficiently sharp.
12 . The method according to claim 1 , characterized in that the comparison of the blur score value and the blur score threshold value is implemented by generating a score ratio by dividing the blur score value with the blur score threshold value, wherein score ratios having a value greater than 1 or not less than 1 indicate that the input image is sufficiently sharp.
13 . A data processing system comprising means for carrying out the steps of the method according to claim 1 .
14 . A non-transitory computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of claim 1 .
15 . A non-transitory computer readable medium comprising instructions which, when executed by a computer, cause the computer to:
provide a blur score threshold value,
input the input image to an image processing system,
generate a two-dimensional frequency spectrum of the input image,
generate a one-dimensional frequency spectrum from the two-dimensional frequency spectrum,
fit a straight line on the one-dimensional frequency spectrum, and
determine a blur score value based on a residual of the fitting, and considering the input image as sufficiently sharp based on a comparison of the blur score value with the blur score threshold value, wherein the comparison is implemented by generating a score ratio by dividing the blur score value with the blur score threshold value, wherein score ratios having a value greater than 1 or not less than 1 indicate that the input image is sufficiently sharp.
16 . The method of claim 5 , wherein a frequency corresponding to a zero frequency is arranged in a centre of the two-dimensional frequency spectrum and frequencies increase radially outwards from the centre.
17 . The method of claim 8 , wherein calculating the blur score value comprises:
calculating the blur score value by generating a further two-dimensional frequency spectrum of each blurred image, by generating a further one-dimensional frequency spectrum of the further two-dimensional frequency spectrum, and by fitting a further straight line on the further one-dimensional frequency spectrum, wherein the blur score value is determined based on a residual of the fitting.