摘要
以数理统计理论和Dempster Shafer证据理论为手段 ,提出了充分利用末制导图像信息处理和识别过程中产生的中间信息和先验知识 ,从而提高目标识别性能的方法 ,该方法包括证据可用性的判断 ,证据对识别结果的支持度的计算模型以及各证据对识别结果支持度的融合计算 .在对实际复杂背景下的红外图像识别实验中 ,该方法可有效降低目标识别中的虚警率和漏警率 ,虚警率和漏警率分别从 0 .0 31和 0 .0 85降低至 0 .0 0 8和 0 .0 38,分别降低了约 3/ 4和 1/ 2 ,表明该方法在提高目标识别性能上是可行和有效的 .
A method was proposed to improve the target recognition quality by combining the physical constrains or prior knowledge as evidences into target recognition within the frame of mathematical statistic theory and Dempster Shafer′s evidence theory. In this method, the usability of the evidences was appraised with Kolmogorov Smirnov test method and the different computation models to compute the belief value to classifier′s result corresponding to the different evidence types proposed. The method was tested on the real infrared images sequences with complex background, reducing the false alarm from 0.031 to 0.008 and miss alarm from 0.085 to 0.038 respective, that is, 3/4 and 1/2 times respectively. The result indicates that the proposed method for improving the recognition performance is feasible and effective.
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2005年第3期20-23,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金重点项目资助 (6 0 135 0 2 0 ) .