Image prediction

A computer-readable storage medium comprising computer-executable instructions that, when executed by a processor, causes the processor to perform operations comprising:

• selecting a topic for prediction;
• considering a plurality of training points pertaining to the topic for prediction;
• determining a mean of the plurality of training points on a shape space, wherein each training point of the plurality of training points corresponds to a square root velocity representation of a parameterized curve of a plurality of parameterized curves derived from a plurality of N-dimensional feature vectors, each N-dimensional feature vector corresponding to a training image of a plurality of training images in a training data set;
• warping the plurality of training points from the shape space onto a tangent space defined at the mean;
• performing dimensionality reduction by retaining eigenvectors with eigenvalues greater than 0.1;
• warping a test point onto the tangent space corresponding to the topic for selected for prediction;
• performing dimensionality reduction using a principal component analysis matrix learnt on the tangent space;
• considering a nearest training neighbor based upon a distance of the eigenvectors;
• and
• predicting a test image based upon a test parameterized curve corresponding to the nearest training neighbor.

A system comprising:

• a processor;
• and
• a memory that stores instructions that, when executed by the processor, cause the processor to perform operations comprising
• selecting a topic for prediction,
• considering a plurality of training points pertaining to the topic for prediction,
• determining a mean of the plurality of training points on a shape space, wherein each training point of the plurality of training points corresponds to a square root velocity representation of a parameterized curve of a plurality of parameterized curves derived from a plurality of N-dimensional feature vectors, each N-dimensional feature vector corresponding to a training image of a plurality of training images in a training data set,
• warping the plurality of training points from the shape space onto a tangent space defined at the mean,
• performing dimensionality reduction by retaining eigenvectors with eigenvalues greater than 0.1,
• warping a test point onto the tangent space corresponding to the topic for selected for prediction,
• performing dimensionality reduction using a principal component analysis matrix learnt on the tangent space,
• considering a nearest training neighbor based upon a distance of the eigenvectors, and
• predicting a test image based upon a test parameterized curve corresponding to the nearest training neighbor.