Stream processing of moment invariants for real-time classifiers

This paper introduces a general purpose graphics processing unit (GPGPU) stream processing implementation of moment invariants using an integral image or summed area table approach. Summed area tables have been used to help attain real-time performance for some classifier systems, however due to the computational complexity of moment invariants, a high throughput computational platform is required to obtain real-time processing. The stream programming algorithm is presented and its performance is evaluated and compared with alternate CPU based approaches. The significant performance gains means that moment invariant classifiers can be implemented for real-time performance on a GPGPU that would not be possible on current CPU platforms.