Images Transfer Based On Android Phone And Deblurring Algorithm Research

Intelligent Transportation System Platform (ITS Platform) development in the21st century, used a lot of advanced theory and technology, the next generationtransportation system development. Using information technology to collect, process, manage traffic information is an important part of the system, in which the specific requirements of the acquisition scene of the accident, driving car pictures, on the occurrence of blurred scene, license plate number blurred image recovery.In the context of these developments, this paper in order to achieve intelligent transportation systems that require specific functions for the purpose, with thecurrent research environment and technology trends, can be put to use from theperspective of the image uploaded on Android devices as well as Motion BlurAlgorithm two conducted in-depth research.Image upload in the Android terminal, the data presented by querying the popularity of the Android platform and version changes. Using the server, client collaborative approach to complete the picture upload programs to use Servlet fileupload size limit exceeded; using HTTP subclass as the main picture upload method. Upload pictures automatically generated in the database simultaneously record, store the uploaded file.In the Motion Blur algorithm study, the algorithm development history, reading well-known literature and compare a variety of go fuzzy algorithm. For blinddeconvolution, non-blind deconvolution are described separately, with emphasison the blind deconvolution suggest improvements. To estimate the exact blur trajectory, before the estimated characteristics required to obtain a clear image: first image through bilateral filter to remove noise, by the impact of the filter edge enhanced image, and then cut to obtain the threshold value to filter gradientmap the noise generated in the previous step; then improved fuzzy trajectory estimation step, the non-blind deconvolution added to the phase trajectory estimation, using the non-blind deconvolution each obtained results are substituted i nto the next cycle as the beginning of trajectory estimation values can be obtained so that a more accurate blur trajectory. In the non-blind deconvolution algorithm of choice, more consistent use of super-fit curve model Laplace clear image restoration.And improvement in the results before the algorithm comparison, the proposed method significantly reduces the ring artifacts, restoring a more significant image detail. An objective evaluation of the comparison, the proposed method significantly improved SNR and PSNR compared to the original algorithm. Experimental results show that the proposed algorithm treatment effect compared with theoriginal algorithm is obviously improved.