BLUR DETECTION WITH LOCAL SHARPNESS MAP

• US 20120162527A1
• Filed: 12/22/2010
• Published: 06/28/2012
• Est. Priority Date: 12/22/2010
• Status: Active Grant

• Current Assignee: Project Giants, LLC
• Original Assignee: Tektronix Incorporated (Fortive Corp.)
• Inventors: BAKER, DANIEL G.
• Granted Patent: US 8,754,988 B2
• Time in Patent Office: Days
• US Class Current: 348/607
• CPC Class Codes: G06T 2207/30168 Image quality inspection
• CPC Class Codes: G06T 7/0002 Inspection of images, e.g. flaw detection
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: H04N 17/004 for digital television systems
• CPC Class Codes: H04N 5/14 Picture signal circuitry for video frequency region cameras or camera modules comprising electronic image sensors, or control thereof H04N23/00
• CPC Class Codes: H04N 5/21 Circuitry for suppressing or minimising disturbance, e.g. moiré or halo

Abstract

A single-ended blur detection probe and method with a local sharpness map for analyzing a video image sequence uses two sets of edge filters, one for “fast edges” and the other for “slow edges.” Each set of edge filters includes a horizontal bandpass filter, a vertical bandpass filter and a pair of orthogonal diagonal filters where the frequency response of the fast edge filters overlap the frequency response of the slow edge filters. The video image sequence is input to each filter of each set, and the output absolute values are combined with weighting factors to produce a slow edge weighted sum array and a fast edge weighted sum arra. The respective weighted sum arrays are then decimated to produce a slow edge decimated array and a fast edge decimated array. The ratio of the maximum difference value between the decimated arrays and the maximum value from the fast edge decimated array, weighted by an appropriate factor, produces a localized maximum sharpness value, the log of which produces a dimensionless blur value.

Claims

1. A probe for determining image blurriness comprising:

a first set of edge filters for detecting fast edges within an input image to produce a plurality of fast edge filter outputs;

a second set of edge filters for detecting slow edges within the input image to produce a plurality of slow edge filter outputs;

means for combining the fast edge filter outputs to produce a fast edge filter output array;

means for combining the slow edge filter outputs to produce a slow edge filter output array;

means for decimating the fast edge and the slow edge filter output arrays separately to produce decimated fast edge filter array and a decimated slow edge filter array; and

means for processing the respective decimated fast edge and slow edge filter arrays to generate a localized maximum sharpness value as a measure of the blurriness of the input image.

2. The probe as recited in claim 1 wherein the first and second sets of edge filters each comprise:

a horizontal edge filter having the input image as an input and producing a horizontal filter output;

a vertical edge filter having the input image as an input and producing a vertical filter output;

a first diagonal edge filter having the input image as an input and producing a first diagonal filter output; and

a second diagonal edge filter orthogonal to the first diagonal edge filter having the input image as an input and producing a second diagonal filter output.

3. The probe as recited in claim 2 wherein each edge filter comprises a filter having a bandpass characteristic along a principal direction and a lowpass filter characteristic along an orthogonal direction to the principal direction, the bandpass frequencies for the first set of edge filters overlapping the bandpass frequencies for the second set of edge filters.

4. The probe as recited in claim 3 wherein each edge filter comprises a filter having a filter kernel of 0s, 1s and 2s that define the bandpass characteristic along the principal direction and the lowpass filter characteristic along the orthogonal direction.

5. The probe as recited in claim 2 wherein the fast edge filter outputs combining means and the slow edge filter outputs combining means each comprise:

a first summation block for combining the horizontal edge filter output with the vertical edge filter output to produce a first edge filter summation output;

a second summation block for combining the first diagonal edge filter output with the second diagonal edge filter output to produce a second edge filter summation output;

means for weighting each of the first and second edge filter summation outputs to produce respective weighted sum edge filter outputs; and

a third summation block for combining the respective weighted sum edge filter outputs to produce the fast edge and slow edge filter arrays respectively.

6. The probe as recited in claim 1 wherein the decimating means comprises:

a lowpass filter for each of the fast edge and slow edge filter arrays having the respective edge filter array as an input to produce respective filtered edge filter arrays; and

a decimator for decimating each of the filtered edge filter arrays to produce the respective decimated edge filter arrays.

7. The probe as recited in claim 6 wherein the decimating means further comprises a crop block for cropping the edges of the respective fast edge and slow edge filter arrays prior to input to the respective lowpass filters.

8. The probe as recited in claim 1 wherein the processing means comprises:

means for combining the decimated fast edge and decimated slow edge filter arrays to produce a difference filter array;

means for generating a ratio between the difference filter array and the decimated fast edge filter array to produce the localized maximized sharpness value.

9. The probe as recited in claim 8 wherein the ratio generating means comprises:

means for obtaining the maximums of the decimated fast edge filter array and the difference filter array;

means for dividing the maximum difference filter array by the maximum fast edge filter array to produce the localized maximized sharpness value.

10. The probe as recited in claim 9 wherein the ratio generating means further comprises means for multiplying the output from the dividing means by a sharpness factor to produce the localized maximized sharpness value.

11. The probe as recited in claim 10 wherein the ratio generating means further comprises means for obtaining the log of the output from the multiplying means to produce the localized maximized sharpness value as a dimensionless value.

12. The probe as recited in claim 2 further comprising means for disabling the horizontal and vertical edge filter outputs of each of the first and second sets of edge filters so the edge filter outputs from the first and second sets of edge filters reflect only the filter outputs from the respective diagonal edge filter pairs.

13. The probe as recited in claim 1 wherein the processing means comprises:

a difference block having as inputs the fast edge and slow edge decimated filter output arrays to produce a difference filter output array;

means for generating from the difference filter array and a sharpness ratio localized markers for superimposing on the input image to indicate where the sharpest edges occur within the input image.

14. The probe as recited in claim 13 wherein the generating means comprises:

a threshold generator having the difference filter output array and the sharpness ratio as inputs and producing a threshold value as an output;

a comparator having the difference filter output array and the threshold value as inputs and producing the localized markers as output when the difference filter output array exceeds the threshold value.

15. The probe as recited in claim 2 further comprising:

means for combining the horizontal and vertical edge filter outputs for the first set of edge filters to produce a first fast edge filter output array;

means for combining the first and second diagonal edge filter outputs for the first set of edge filters to produce a second fast edge filter output array;

means for decimating the first and second fast edge filter output arrays to produce decimated first and second fast edge filter output arrays; and

a combination block for comparing the ratio of the first and second fast edge filters with a tiling threshold to produce a tiling flag indicative of block artifacts within the input image.

16. The probe as recited in claim 15 further comprising means for disabling the horizontal and vertical edge filter outputs of each of the first and second sets of edge filters in response to the tiling flag so the edge filter outputs from the first and second sets of edge filters reflect only the filter outputs from the respective diagonal edge filter pairs.

17. The probe as recited in claim 1 further comprising:

a comparison block for comparing the localized maximum sharpness value with a sharpness threshold; and

means for generating an up conversion flag when the sharpness threshold is less than the localized maximum sharpness value for a predetermined period of time, indicating the input image was upconverted from a lower resolution image.

18. A method of determining image blurriness comprising the steps of:

filtering an input image with a first set of edge filters for detecting fast edges in the input image to produce a plurality of fast edge filter outputs;

filtering the input image with a second set of edge filters for detecting slow edges in the input image to produce a plurality of slow edge filter outputs;

combining the fast edge filter outputs to produce a fast edge filter output array and the slow edge filter outputs to produce a slow edge filter output array;

decimating the fast edge and the slow edge filter output arrays separately to produce a decimated fast edge filter array and a decimated slow edge filter array; and

processing the respective decimated fast edge and slow edge filter arrays to generate a localized maximum sharpness value as a measure of the blurriness of the input image.

19. The method as recited in claim 18 further comprising the steps of:

obtaining a difference between the fast edge and slow edge decimated filter output arrays to produce a difference filter output array;

generating from the difference filter output array and a sharpness ratio localized markers for superimposing on the input image to indicate where the sharpest edges occur within the input image.

20. The method as recited in claim 18 further comprising the steps of:

combining horizontal and vertical edge filter outputs for the first set of edge filters to produce a first fast edge filter output array;

combining first and second diagonal edge filter outputs for the first set of edge filters to produce a second fast edge filter output array;

decimating the first and second fast edge filter output arrays to produce decimated first and second fast edge filter output arrays; and

comparing a ratio of the first and second fast edge filters with a tiling threshold to produce a tiling flag indicative of block artifacts within the input image.