Identifying gray regions for auto white balancing
Examples of the disclosure accurately define a gray region for automatic white balancing. Ratios corresponding to possible gray areas are white balanced before color distance metrics are calculated with respect to white points under different light sources. The color distance metrics of each point in the image data are compared with a predetermined threshold value to identify a gray region. Gray points within a captured image are identified using the gray region and input to any AWB method to accurately white balance the captured image.
1. A system, comprising:
a sensor configuration component that when executed causes at least one processor to define a gray region for an image sensor through:
accessing image data,
identifying a red/green ratio and a blue/green ratio for a point in the image data,
identifying white points in the image data, and
white balancing the point in the image data using the white points to produce white-balanced data, wherein said white balancing is performed through application of one or more white-balancing gains to at least one of the red/green ratio or the blue/green ratio associated with the point; and
a camera component that when executed causes the at least one processor to identify one or more gray points in the image data based on the white-balanced data.
2. The system of claim 1 , wherein the sensor configuration component is further configured for interpolating the at least one of the white points to form a curve associated with the image data.
3. The system of claim 1 , wherein the sensor configuration component is further configured for calculating a color distance for the point after said white balancing is performed, wherein the color distance comprises a distance of the white-balanced data associated with the point to a white point in the image data.
4. The system of claim 3 , wherein the color distance is calculated based on the red/green ratio and the blue/green ratio.
5. The system of claim 1 , wherein said identification of the one or more gray points is performed by curve fitting a line segment to the image data and determining a distance of the one or more gray points to the line segment.
6. The system of claim 1 wherein the sensor configuration component causes the at least one processor to:
compute a color distance between the point and at least one of the white points, and
identify a gray point based on the computed color distance being less than a threshold color distance value.
7. The system of claim 1 , wherein the camera component further averages the red/green and blue/green ratios in said white balancing of the accessed data to produce the white-balanced data.
8. The system of claim 7 , wherein the one or more gray points are associated with a constraint value that exceeds the one or more white-balancing gains.
9. A method for defining a gray region for white balancing, the method comprising:
accessing red/green or blue/green ratios corresponding to a point in an image;
white-balancing the image through applying one or more white-balancing gains to the red/green or blue/green ratios corresponding to the point to generate white-balanced data;
calculating one or more color distances to one or more white points based on the white-balanced data; and
defining a gray region in the image based on the calculated one or more color distances.
10. The method of claim 9 , wherein defining the gray region in the image comprises comparing the calculated one or more color distances to a threshold value associated with at least one of the one or more white points.
11. The method of claim 9 , further comprising:
interpolating the one or more white points to form a curve associated with the image data; and
curve fitting the one or more white points to the image data.
12. The method of claim 9 , further comprising wherein the calculated one or more color distances comprises a distance of the white-balanced data associated with the point to a white point in the image data.
13. The method of claim 9 , wherein the calculated one or more color distances are calculated based on both the red/green and blue/green ratios.
14. The method of claim 9 , further comprising representing the gray region with two sets of line segments defining upper and lower boundaries of the gray region.
15. The method of claim 14 , wherein defining the gray region comprises using curve fitting for the upper and lower boundaries using at least two polynomials and a range of red/green values for points in the image data.
16. One or more computer-storage media in the form of machine-readable memory embodied with executable instructions for identifying gray regions in image data, comprising:
identifying one or more white points in the image data;
white balancing other points of the image data based on the one or more white points;
determining an upper limit boundary and a lower limit boundary in the image data based on the white-balanced other points;
curve fitting at least one of the upper limit boundary or the lower limit boundary to define the gray region; and
identifying one or more points in the image data being within the gray region based on said curve-fitted upper limit boundary or lower limit boundary.
17. The media of claim 16 , wherein at least one of the upper limit boundary or the lower limit boundary are represented as a line segment.
18. The media of claim 16 , wherein at least one of the upper limit boundary or the lower limit boundary are represented as a polynomial.
19. The media of claim 16 , wherein the gray region is defined through calculating a range of blue/green values associated with the one or more points in the image data as a function of the upper boundary limit or the lower boundary limit.
20. The system of claim 16 , wherein said white balancing of the other points of the image data comprises applying one or more white-balancing gains to blue/green or red/green values of the other points to produce associated white-balance data.