Parallel Rapid Computing Development And Research (including Pc), Image Edge Detection Based On The Ibm Machine

After IBM z900 was introduced into UESTC, I hoped to get some achievements of parallel computing based on mainframe through my research and analysis. Because of the corresponding computing modules, such as MPI, C compiler, etc. in mainframe, I utilized multiple LPA's to accomplish the brink computing of high resolution bitmap and realized the goal successfully. In this process, I also got lost of operation experience about mainframe.Parallel processing technique deals with many aspects. It includes hardware technique, parallel architectures, parallel operating system, parallel languages, parallel compiler, parallel software, parallel algorithm, etc. But the current international situation is that the development of parallel algorithm is far more behind other architectures. The most dominating appearances are the lacks of corresponding software, the rational distributing of tasks, the communications and the synchronization.Because the number of data in image disposal processing is very large, and the existence of many convolution operations and matrix multiplication operations in all kinds of algorithms, it is possible that parallel algorithm design and application can be used in image disposal processing.Edge detection is always one of the most classical studying projects of computer vision and image processing field. The fist step of image analysis and understanding is edge detection. The goal of edge detection is to recover information about shapes and reflectance or transmittance in an image. It is one of the fundamental steps in image processing , mage analysis , image patter recognition , and computer vision , as well as in human vision. The correctness and reliability of its results affect directly the comprehension which machine system make for objective world.Firstly, the classic methods of image edge detection are introduced, including differential coefficient which are Roberts image edge detection, Sobel edge detection, and Prewitt, Robinson, Kirsch image edge detection, and second-order differential coefficient which are Laplacian edge detection and LOG edge detection. Finally analyzed the experiment results of these methods application on a picture.Secondly, according to a case of Kirsch-filter, the method of reducing computing step of linear filter is introduced and discussed. The complexity of improved method is half of the rural one. It proves that the improving effect is obviously.At last, parallel algorithm design is researched and applied to image edge detection, some good results are gained correspondingly on a simulation cluster system. At the same time, by using theoretical analysis and simulated computation, with reference to the acceleration rates and parallel efficiencies, the improvement of algorithm design method is discussed.