Foreground object detection from multiple images
A method for determining a salient region of an image is disclosed. For a plurality of different saliency cue functions, a single saliency value is calculated for each pixel in a plurality of adjacent pixels in an image using the saliency cue function, wherein one of the saliency cue functions is based on whether the pixel is in a region of the image whose colors contrast with the region's background and another of the saliency cue functions is based on a foreground and background color models of the image. A classifier is used to calculate a combined single saliency value for each pixel based on the single saliency values for the pixel. The salient region of the pixels is determined with a subwindow search based on the combined single saliency values.
1. A computer-implemented method, comprising:
identifying sample regions that each include a plurality of adjacent pixels of an image;
calculating, for pixels in the sample regions and by one or more processors, a foreground saliency value of the pixel based, at least in part, on a conditional probability that the pixel is a foreground pixel based on a foreground color model and a conditional probability that the pixel is a background pixel based on a background color model;
using a classifier to calculate a respective combined saliency value for each pixel based on the foreground saliency value for the pixel and a saliency value for at least one different saliency cue function wherein the classifier is trained on training image and corresponding saliency values and wherein the respective combined saliency value for each of the training images comprise saliency values for a plurality of respective pixels in the training image; and
determining, by one or more processors, a salient region of the pixels with a subwindow search based on the respective combined saliency values in a rectangle that maximizes a sum of combined saliency values of pixels in the rectangle and a complement of the combined saliency values of the pixels in the rectangle.
2. The method of claim 1 , wherein the classifier is a support vector machine.
3. The method of claim 1 , wherein one of the saliency cue functions is based on whether the pixel is in a region of the pixels whose colors contrast with the region's background and another of the saliency cue functions is based on foreground and background Gaussian mixture models.
4. The method of claim 3 , further comprising segmenting the plurality of adjacent pixels into super-pixels and using a super-pixel algorithm process to determine the region including the pixel that has the highest difference with its background.
5. A non-transitory computer storage medium encoded with a computer program, the program comprising instructions that when executed by one or more data processing apparatus cause the one or more data processing apparatus to perform operations comprising:
identifying sample regions that each include a plurality of adjacent pixels of an image;
calculating, for pixels in the sample regions, a foreground saliency value of the pixel based, at least in part, on a conditional probability that the pixel is a foreground pixel based on a foreground color model and a conditional probability that the pixel is a background pixel based on a background color model;
using a classifier to calculate a respective combined saliency value for each pixel based on the foreground saliency value for the pixel and a saliency value for at least one different saliency cue function wherein the classifier is trained on training images and corresponding saliency values, and wherein the respective combined saliency value for each of the training images comprise saliency values for a plurality of respective pixels in the training image; and
determining a salient region of the pixels with a subwindow search based on the respective combined saliency values in a rectangle that maximizes a sum of combined saliency values of pixels in the rectangle and a complement of the combined saliency values of the pixels in the rectangle.
6. The computer storage medium of claim 5 , wherein the classifier is a support vector machine.
7. The computer storage medium of claim 5 , wherein one of the saliency cue functions is based on whether the pixel is in a region of the pixels whose colors contrast with the region's background and another of the saliency cue functions is based on foreground and background Gaussian mixture models.
8. The computer storage medium of claim 7 , wherein the instructions cause the one or more data processing apparatus to perform operations further comprising segmenting the plurality of adjacent pixels into super-pixels and using a super-pixel algorithm process to determine the region including the pixel that has the highest difference with its background.
9. A system comprising:
a data storage device storing instructions; and
one or more data processing apparatus that interact with the data storage device and execute instructions that cause the one or more data processing apparatus to perform operation comprising:
identifying sample regions that each include a plurality of adjacent pixels of an image;
calculating, for pixels in the sample regions, a foreground saliency value of the pixel based, at least in part, on a conditional probability that the pixel is a foreground pixel based on a foreground color model and a conditional probability that the pixel is a background pixel based on a background color model;
using a classifier to calculate a respective combined saliency value for each pixel based on the foreground saliency value for the pixel and a saliency value for at least one different saliency cue function wherein the classifier is trained on training images and corresponding saliency values, and wherein the respective combined saliency value for each of the training images comprise saliency values for a plurality of respective pixels in the training image; and
determining a salient region of the pixels with a subwindow search based on the respective combined saliency values in a rectangle that maximizes a sum of combined saliency values of pixels in the rectangle and a complement of the combined saliency values of the pixels in the rectangle.
10. The system of claim 9 , wherein the classifier is a support vector machine.
11. The system of claim 9 , wherein one of the saliency cue functions is based on whether the pixel is in a region of the pixels whose colors contrast with the region's background and another of the saliency cue functions is based on foreground and background Gaussian mixture models.
12. The system of claim 11 , wherein the instructions cause the one or more data processing apparatus to perform operations further comprising segmenting the plurality of adjacent pixels into super-pixels and using a super-pixel algorithm process to determine the region including the pixel that has the highest difference with its background.