Local feature representation for image recognition

A non-transient computer program product having instructions encoded thereon that, when executed by one or more processors, causes a process to be carried out, the process comprising:

• dividing a digital image into a plurality of image patches;
• generating an image patch vector for each of the image patches;
• comparing each image patch vector to Gaussian mixture components of a Gaussian mixture model, thereby generating a similarity score for each image patch vector;
• generating a final image feature vector that does not include image patch vectors associated with a similarity score that is below a given threshold;
• and
• categorizing the digital image using the final image feature vector.

A computing system comprising:

• a memory resource configured to store executable instructions;
• and
• a processor configured to execute the instructions to:
• receive a digital image that includes a plurality of image features, wherein the digital image is divided into a plurality of image patches, each of which represents only a portion of the received digital image;
• generate an image patch vector for each of the image patches;
• compare each image path vector to Gaussian mixture components of a Gaussian mixture model, each Gaussian mixture components being a vector, thereby generating a similarity score for each image patch vector;
• eliminating one or more image patch vectors associated with a similarity score that is below a given threshold, thereby establishing a set of remaining image patch vectors;
• and
• generating a final image feature vector using the set of remaining image patch vectors.