针对传统ViBe算法不能及时反映场景变化,动态场景适应性差等问题,提出一种改进的ViBe算法.改进内容包括:采用随机选取背景样本和24邻域法获取初始背景,可以加速"鬼影"消融;结合大津法(OTSU)和均匀性度量法的平均自适应阈值计算方法,可以提高算法对树叶晃动、水波纹和光照变化等环境的适应性,最大限度保留有效像素;更新阶段引入自适应更新因子,可以有效减少被误判的概率,从而增强算法的鲁棒性;最后通过形态学处理和滤波使目标更加完整.采用标准数据集视频对改进算法进行了测试和对比分析,改进算法相对于KDE算法、GMM算法和传统ViBe算法各项指标均有大幅度提高,精确度分别提高30.44％、40.72％和20.95％,错分比分别降低了43.28％、40.59％和29.43％.
The traditional iSCSI transfers file by fixing the number of TCP connections between both ends,which severely limits the performance of the system.This paper presents a multi-link adaptive adjustment mechanism which can be based on network status,using variants of parabolic interpolation method to automatically change the number of TCP connections to optimize system performance.Compared with the traditional iSCSI system,experiments show that this mechanism can effectively determine the optimal value and improve the throughput of the system.
Aimed at the problems of the iSCSI bottleneck between the I/O of disk and the network speed,the improved iSCSI storage architecture is presented.This strategy can dynamic select storage strategy according to the size of request packets and eliminate the redundant overwrites.Thus the latency caused by multiple reads to same storage sector are reduced in the parallel algorithm of target.The experiment shows that the strategy has shorter response time and higher throughput to improve the iSCSI performance.
