IMAGE CLASSIFICATION
Images are classified as photos (e.g., natural photographs) or graphics (e.g., cartoons, synthetically generated images), such that when searched (online) with a filter, an image database returns images corresponding to the filter criteria (e.g., either photos or graphics will be returned). A set of image statistics pertaining to various visual cues (e.g., color, texture, shape) are identified in classifying the images. These image statistics, combined with pre-tagged image metadata defining an image as either a graphic or a photo, may be used to train a boosting decision tree. The trained boosting decision tree may be used to classify additional images as graphics or photos based on image statistics determined for the additional images.
1 . A method, comprising:
classifying an image as at least one of a photo or a graphics image based upon:
a difference between a first energy value of a lower frequency color band and a second energy value of a higher frequency color band; and
at least one of:
a total number of edges in the image; or
an average length of edges in the image.
2 . The method of claim 1 , the classifying performed using a boosting decision tree.
3 . The method of claim 2 , the boosting decision tree trained using one or more classifying features of the image.
4 . The method of claim 3 , at least some of the one or more classifying features determined using at least one of the total number of edges in the image or the average length of edges in the image.
5 . The method of claim 1 , comprising using an edge detector to determine at least one of the total number of edges in the image or the average length of edges in the image.
6 . The method of claim 1 , comprising using a canny edge detector to determine at least one of the total number of edges in the image or the average length of edges in the image.
7 . The method of claim 1 , the classifying performed using a gradient magnitude-orientation histogram.
8 . The method of claim 7 , comprising determining an entropy value associated with the gradient magnitude-orientation histogram.
9 . The method of claim 1 , the classifying based upon a spatial correlogram of gray level pixels of the image.
10 . A computer readable medium comprising instructions that when executed, perform a method comprising:
classifying an image as at least one of a photo or a graphics image, comprising measuring a difference between a first energy value of a lower frequency color band and a second energy value of a higher frequency color band.
11 . The computer readable medium of claim 10 , the classifying performed using a gradient magnitude-orientation histogram.
12 . The computer readable medium of claim 10 , the classifying performed using a boosting decision tree.
13 . The computer readable medium of claim 12 , the boosting decision tree trained using one or more classifying features of the image.
14 . The computer readable medium of claim 11 , the method comprising determining an entropy value associated with the gradient magnitude-orientation histogram.
15 . A system comprising:
one or more processing units; and
memory comprising instructions that when executed by at least one of the one or more processing units perform a method comprising:
classifying an image as at least one of a photo or a graphics image based upon at least one of:
a total number of edges in the image; or
an average length of edges in the image.
16 . The system of claim 15 , the classifying performed using a boosting decision tree.
17 . The system of claim 16 , the boosting decision tree trained using one or more classifying features of the image.
18 . The system of claim 17 , at least some of the one or more classifying features determined using at least one of the total number of edges in the image or the average length of edges in the image.
19 . The system of claim 15 , the method comprising using an edge detector to determine at least one of the total number of edges in the image or the average length of edges in the image.
20 . The system of claim 15 , the method comprising using a canny edge detector to determine at least one of the total number of edges in the image or the average length of edges in the image.