首页> 美国政府科技报告 >Novel Graph Searching Methods for Image Analysis.
【24h】

Novel Graph Searching Methods for Image Analysis.

机译:一种新的图像分析图搜索方法。

获取原文

摘要

A generalization of the Artificial search algorithm, SSS*, is given, allowing it to handle incomplete and inconsistent data. A parallel formulation of the generalized SSS* is developed and implemented on a BBN Butterfly Plus parallel processor. In order to make the parallel version of SSS* as efficient as possible, a concurrent heap data structure (based on the work of Nageshwara and Kumar) is utilized. Theoretical speedup due to the use of the concurrent heap is shown. An automatic tree generator is developed to allow the testing of parallel SSS* on multiple trees. Empirical results are given for testing SSS* on 1 to 40 processors. Speedup proportional to the number of processors for up to 24 processors is achieved depending on the size of the search space. A procedure for achieving rapid output of solutions by using a tradeoff between the estimated complexity and quality of solutions was developed. Finally, future directions for research are presented. (RH)

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号