Digital document image classification method using GPU-based parallel programming with Cuda
The increasing use of document images creates another need to classify documentrnimages into their type automatically. Some feature extraction methods have beenrnproposed in order to be able to recognize the document images and classify it into itsrntype. One of the simple and effective feature extraction method is by using binaryrnmorphological erosion with 9 intersection types that is proposed by Neves at al., 2007.rnHowever this method is turned to be not applicable because of time processing. Thisrnresearch focuses on the use of multithreading in multi-core processor with POSIXrnThreads (Pthreads) and multithreading GPU-based with NVIDIA CUDA technologyrnto speed up processing time. Result shows that with local processing, thernclassification accuracy can be improved. Regarding processing time, thernimplementation of POSIX Threads in dual core processor can increase the processingrnspeed around 2 scale factor, in the other hand, the implementation of CUDA canrnincrease the processing speed around 100 scale factor compare with single threadrnprogram and 60 scale factor compare with multithreading in dual-core processor.
|B00947||(wh)||Available but not for loan - Missing|
No other version available