Detection of objects in an image using self similarities

Abstract

An image processor ( 10 ) has a window selector for choosing a detection window within the image, and a self similarity computation part ( 40 ) for determining self-similarity information for a group of the pixels in any part of the detection window, to represent an amount of self-similarity of that group to other groups in any other part of the detector window, and for repeating the determination for groups in all parts of the detection window, to generate a global self similarity descriptor for the detection window. A classifier ( 50 ) is used for classifying whether an object is present based on the global self-similarity descriptor. By using global self-similarity rather than local similarities more information is captured which can lead to better classification. In particular, it helps enable recognition of more distant self-similarities inherent in the object, and self-similarities present at any scale.

1 . An image processor ( 10 ) for detection of an object in an image or sequence of images, each image being formed of pixels, and the image processor comprising:
 a window selector for choosing a detection window within the image,   a self similarity computation part ( 40 ) for determining self-similarity information for a group of the pixels in any part of the detection window, to represent an amount of self-similarity of that group to other groups in any other part of the detector window, and for repeating the determination for groups in all parts of the detection window, to generate a global self similarity descriptor for the chosen detection window, and   a classifier ( 50 ) for classifying whether the object is present in the detection window of the image from the global self-similarity descriptor for that detection window.   
     
     
         2 . The image processor of  claim 1 , the self similarity information comprising an amount of self-similarity of colours of pixels of the group. 
     
     
         3 . The image processor of  claim 1 , having a part ( 42 ) arranged to determine a distribution of colours of the pixels of the groups, and the self similarity information comprising an amount of self-similarity of the colour distributions. 
     
     
         4 . The image processor of  claim 1 , having a part ( 30 ) for determining gradient information by determining a distribution of intensity gradients in a cell of pixels, and inputting such gradient information for cells over all parts of the detection window to the classifier, the classifier additionally being arranged to use the gradient information to classify whether the object is present. 
     
     
         5 . The image processor of  claim 1 , having a part arranged to determine a flow of the groups in terms of motion vectors of the pixels of the groups over successive images in a sequence of images, and the self-similarity information comprising an amount of self-similarity of the flow. 
     
     
         6 . The image processor of  claim 1 , the self-similarity computation part having a histogram generator ( 44 ) arranged to determine a histogram of values for a feature of pixels in the group, by using interpolation. 
     
     
         7 . The image processor of  claim 6 , the self similarity computation part having a part ( 46 ) arranged to determine similarities between histograms for different groups of pixels in the detection window by a histogram intersection. 
     
     
         8 . The image processor of  claim 1 , comprising a motion detection part ( 70 ) for detecting motion vectors for parts of the image, and the classifier part being arranged to classify based also on the motion vectors of parts in the detection window. 
     
     
         9 . The image processor of  claim 1 , having a combiner part ( 60 ) for combining the similarity information and the distributions of intensity gradients before input to the classifier. 
     
     
         10 . A method of using an image processor for detection of an object in an image or sequence of images, each image being formed of pixels, and the method having the steps of:
 choosing ( 100 ,  130 ) a detection window within the image,   using the image processor to determine ( 110 ) self-similarity information for a group of the pixels in any part of the detection window, to represent an amount of self-similarity of that group to other groups in any other part of the detector window,   repeating the determination ( 130 ) for groups in all parts of the detection window, to generate a global self similarity descriptor for the chosen detection window, and   classifying ( 140 ) whether the object is present in the detection window of the image from the global self-similarity descriptor for that detection window.   
     
     
         11 . The method of  claim 10 , having the step of determining a distribution of colours of the pixels of the groups, and the self similarity information comprising an amount of self-similarity of the colour distributions. 
     
     
         12 . The method of  claim 10 , having the steps of determining gradient information by determining a distribution of intensity gradients in a cell of pixels, and determining such gradient information for cells over all parts of the detection window, and the classifying step additionally using the gradient information to classify whether the object is present. 
     
     
         13 . The method of  claim 10 , having the step of determining a flow of the groups in terms of motion vectors of the pixels of the groups over successive images in a sequence of images, and the self-similarity information comprising an amount of self-similarity of the flow. 
     
     
         14 . A program on a computer readable medium and having instructions which when executed by a computer cause the computer to carry out the method of  claim 10 . 
     
     
         15 . An integrated circuit having the image processor of  claim 1 .