Design Of Memory Scheduling Algorithm For File Traversal And Its Application In Face Recognition System

The traditional large data files are generally stored in a disk or CD etc secondary storage medium, the application must transform data from a disk or CD to the memory before computing and processing. A right memory scheduling algorithm can effectively avoid disk thrash, and improve software efficiency. However, under the CPU protected mode, only the operating system have the right of memory scheduling, the application software can only rely on the memory scheduling algorithm provided by the operating system to process mass data. When the memory scheduling algorithm provided by the operating system is not suitable for the software page access mode, the efficiency of the software will be discounted.This paper puts forward a memory scheduling solution and mechanism through combining Windows itself memory scheduling strategy in single-machine and network environment, so that the application can be no longer limited to operating system memory scheduling strategy and can free to choose the most suitable memory scheduling algorithm, thus achieved to improve the software effectively.In single-machine environment, the solution can establish a right memory scheduling algorithm for a software according to the software page access mode based on the operating system original memory scheduling strategy, reduce disk literacy times, improve the software efficiency.In network environment, by using the distributed computing technology, the algorithm handle the huge mass data work cutting into multiple small task, and assign the task to a workstation reasonably according to the workstation's memory state, so that can balance the resources of the workstation, and improve the efficiency of the system.