high performance computing on graphics processing units: hgpu.org

Due to their massively parallel hardware design, graphic processors can easily beat ordinary CPUs in applications which involve large amount of data. Considering their great potential, the objective of this paper is to continue previous work and optimize the speed and efficiency of texture and fractal analysis, as used for medical images classification processes for early skin cancer detection. The images are classified based upon Haralick features and fractal geometry, which both proved to be traditionally computationally expensive.