一种Omni-顺序扫描的补充算法

A Complementary Algorithm Based on Omni-Sequential Scan Algorithm

提出了一种Omni＿顺序扫描的补充算法,修改了原Omni＿坐标的结构,使新的索引文件在完成利用三角不等式进行过滤的原有功能之外,对剩余的对象进行了第二级过滤,再次减少了所需进一步比较对象的数目,有效减少了I/O次数,提高了索引算法的效率.文中对500幅RGB位图进行了6组实验,分别对应不同的焦点数和查询半径,新算法较原算法的检索效率的提高率分别为:21.9%,30.3%,19.2%,23.2%,17.3%,18.4%.实验数据验证了算法的有效性.此外,并对实验数据进行了比较分析.

In this paper, a complementary algorithm based on Omni_sequential scan algorithm is proposed. The structure of Omni_coordinate is modified. In addition to implementation of pruning distant calculations through the triangle inequality property, the new sequential scan method can be used to further filter the remained data. Through this step, the number of objects to be accessed is greatly reduced and the number of I/O is reduced accordingly. The efficiency of the new sequential scan algorithm is significantly improved. In this paper, six groups experiments with different foci and query radius are done on five hundred RGB bitmaps. The improved efficiency is 21.9%, 30.3%, 19.2%, 23.2%, 17.3%, 18.4% respectively. The experimental results show the efficiency of the new algorithm. In addition, the results are compared and analyzed with the experiment data.