Abstract: Image Retrieval instruments can aid individuals in making productive utilization of computerized picture accumulations; likewise it has gotten to be basic to discover productive techniques for the retrieval of these pictures. Most picture preparing calculations are innately parallel, so multithreading processors are suitable in such applications. In huge picture databases, picture preparing takes long time for run on a solitary(single core processor) center processor due to single string execution of calculations. GPU is more basic in most picture transforming applications because of multithread execution of calculations, programmability and minimal effort. In this paper we execute shading minutes and surface based picture Retrieve (entropy, standard deviation and neighborhood range) in parallel utilizing CUDA programming model to run on GPUs. These features are connected to inquiry pictures from a database which is like an inquiry picture. We assessed our image retrieve framework utilizing review, exactness, and normal accuracy measures. Exploratory results demonstrated that parallel execution prompted a normal accelerate of 144.67×over the serial execution when running on a NVIDIA GPU GeForce GT610M. Additionally the normal exactness and the normal review of proposed strategy are 61.968% and 55% individually.
Keywords: CBIR, Color moments, CUDA, GPU.
Title: Acceleration of Image Retrieval System Using CUDA Based Parallel Computing On GPU
Author: Shingade Suraj, Patil Ganesh, Bhamare Sameer, Kumar Saurabh
International Journal of Computer Science and Information Technology Research
ISSN 2348-1196 (print), ISSN 2348-120X (online)
Research Publish Journals
