The Research And Implementation Of Image Processing Technology Based On Extreme Learning Machine

The processing and applications of medical image,not only can make full use of existing medical imaging equipment,and greatly improve the level of clinical diagnosis,but also make the basis of medical teaching,training,computer aided clinical surgery by the provision of electronic means of implementation,to provide a solid foundation for medical research and development.In the process of medical treatment,Magnetic Resonance Imaging(MRI)is often used to test the lesion area,MRI images have a direct influence to the accuracy of the diagnosis of doctor,MRI can be applied to the human body of the multiple parts with a multi-angle and multiplanar imaging process,therefore can show the body's tissues more objective and specific to a better observation and diagnosis on the lesion.In this paper,the research work includes the following aspects:This paper expounds the existing way of image super-resolution and the way of image segmentation,introduces the content of Extreme Learning Machine(ELM)and its development and application detailed.In this paper,the Extreme Learning Machine(ELM)is applied to the super-resolution of nuclear magnetic resonance image,and a method which is proposed with the combination of sparse representation and Extreme Learning Machine for the super-resolution of single MRI image,through the establishment of mapping relation model between the high resolution image and the low resolution image to realize the image super-resolution reconstruction.This article uses the real brain imaging data to test the algorithm,this method was verified with general way based on Support Vector Regression(SVR)by compared with faster training speed and better effect of super-resolution.Extreme Learning Machine(ELM)is applied to the segmentation of MRI image,and the use of general Extreme Learning Machine is in different ways,this paper uses a kind method of unsupervised image segmentation.The unsupervised Extreme Learning Machine(US-ELM)and the fuzzy c-means method based on spatial neighborhood information is union to achieve an unsupervised artificial intervention segmentation of MRI image.This paper uses the real brain image data to test the algorithm,and the segmentation effect is analyzed,to prove that it has better segmentation effect.All in all,this paper studies the single MRI image super-resolution method based on the extreme learning machine as well as the image super-resolution method based on unsupervised extreme learning machine.These methods describe the application of extreme learning machine in MRI images,which has a certain significance to the development of medical image processing.