Image filtering
Pixels of an image ( 200 ) are coded as color vectors specified by RGB values. The image ( 200 ) is filtered by calculating new color vectors for the pixels on a pixel by pixel basis. A new color vector for a subject pixel ( 201 ) is calculated from the average of neighbor pixels ( 203 ) in a window ( 202 ) around the subject pixel ( 201 ). A first threshold is calculated from the standard deviation of the color vectors of the neighbor pixels ( 203 ) in the window ( 202 ). A second threshold is calculated from the median maximum difference between the values defining the color vectors of the respective neighbor pixels ( 203 ) in the window ( 202 ) and the standard deviation of the values. Only neighbor pixels ( 203 ) having a color vector that differs from the color vector of the subject pixel ( 201 ) by less than or the same as the first and second thresholds are used in calculation of the new color vector for the subject pixel ( 201 ).
1. A method of filtering an image, the image having a plurality of pixels and a colour of each said pixel being specified by values that define a colour vector for said pixel, the method comprising:
identifying neighbour pixels in a window around a subject said pixel;
determining, for each of the neighbour pixels in the window, a maximum difference between the values defining the colour vector for that neighbour pixel;
calculating a threshold based on the maximum differences and the standard deviation of the values defining the colour vectors for the neighbour pixels in the window;
selecting only those neighbour pixels in the window that have a colour vector that differs from the colour vector of the subject pixel by less than or the same as the threshold; and
determining a new colour vector for the subject pixel based on the colour vectors for the selected neighbour pixels.
2. The method of claim 1 , wherein the threshold is determined based on a median of the maximum differences for the neighbour pixels in the window.
3. The method of claim 1 , wherein the threshold is a scaled product of the maximum differences and a standard deviation of the values defining the colour vectors for the neighbour pixels in the window.
4. The method of claim 1 , further comprising determining the new colour vector for the subject pixel based on an average of the colour vectors for the selected neighbour pixels.
5. The method of claim 1 , further comprising:
calculating another threshold based on the standard deviation of the colour vectors of the neighbour pixels in the window; and
selecting only those neighbour pixels in the window that have a colour vector that differs from the colour vector of the subject pixel by less than or the same as the another threshold.
6. The method of claim 1 , wherein the colour vector of each pixel of the image is defined by RGB values.
7. Computer software for implementing the method of claim 1 when processed by computer processing means.
8. An apparatus for filtering an image, the image having a plurality of pixels and the colour of each said pixel being specified by values that define a colour vector for the pixel, the apparatus comprising a processor for:
identifying neighbour pixels in a window around a subject said pixel;
determining, for each of the neighbour pixels in the window, a maximum difference between the values defining the colour vector for that neighbour pixel;
calculating a threshold based on the maximum differences and the standard deviation of the values defining the colour vectors for the neighbour pixels in the window;
selecting only those neighbour pixels in the window that have a colour vector that differs from the colour vector of the subject pixel by less than or the same as the threshold; and
determining a new colour vector for the subject pixel based the colour vectors for the selected neighbour pixels.
9. The apparatus of claim 8 , wherein the threshold is determined based on a median of the maximum differences for the neighbour pixels in the window.
10. The apparatus of claim 8 , wherein the threshold is a scaled product of the maximum differences and the standard deviation of the values defining the colour vectors for the neighbour pixels in the window.
11. The apparatus of claim 8 , wherein the processor determines the new colour vector for the subject pixel based on the average of the colour vectors for the selected neighbour pixels.
12. The apparatus of claim 8 , wherein the processor:
calculates another threshold based on the standard deviation of the colour vectors of the neighbour pixels in the window; and
selecting only those neighbour pixels in the window that have a colour vector that differs from the colour vector of the subject pixel by less than or the same as the another threshold.
13. The apparatus of claim 8 , wherein the colour vector of each pixel of the image is defined by RGB values.
14. A digital camera comprising the apparatus of claim 8 .