科技通报
科技通報
과기통보
BULLETIN OF SCIENCE AND TECHNOLOGY
2015年
6期
160-162
,共3页
军事禁区%嵌入式%人员监控%特征提取
軍事禁區%嵌入式%人員鑑控%特徵提取
군사금구%감입식%인원감공%특정제취
military restricted zone%embedded%personnel monitoring%feature extraction
通过对军事禁区人员的视觉特征提取提高军事禁区的人员监控和识别能力。提出一种基于嵌入式视觉技术的军事禁区人员监控技术,系统采用4台计算机,3台显示器,虚拟现实设备设计,整个系统的控制是通过连有手控器的计算机来实现军事禁区人员的视觉特征提取,构建视觉特征提取的状态方程和观测方程,对嵌入式视觉特征进行运动属性分析和图像处理,实现人员监控特征分析算法改进。仿真结果表明,该系统能有效实现远距离视频监控和人员特征识别及嵌入式视觉定位,测距误差较低,对军事禁区人员的监控精度提高,对远程监控视频特征的场景恢复性能较好,能有效识别人员的行为和视觉特征,性能优越。
通過對軍事禁區人員的視覺特徵提取提高軍事禁區的人員鑑控和識彆能力。提齣一種基于嵌入式視覺技術的軍事禁區人員鑑控技術,繫統採用4檯計算機,3檯顯示器,虛擬現實設備設計,整箇繫統的控製是通過連有手控器的計算機來實現軍事禁區人員的視覺特徵提取,構建視覺特徵提取的狀態方程和觀測方程,對嵌入式視覺特徵進行運動屬性分析和圖像處理,實現人員鑑控特徵分析算法改進。倣真結果錶明,該繫統能有效實現遠距離視頻鑑控和人員特徵識彆及嵌入式視覺定位,測距誤差較低,對軍事禁區人員的鑑控精度提高,對遠程鑑控視頻特徵的場景恢複性能較好,能有效識彆人員的行為和視覺特徵,性能優越。
통과대군사금구인원적시각특정제취제고군사금구적인원감공화식별능력。제출일충기우감입식시각기술적군사금구인원감공기술,계통채용4태계산궤,3태현시기,허의현실설비설계,정개계통적공제시통과련유수공기적계산궤래실현군사금구인원적시각특정제취,구건시각특정제취적상태방정화관측방정,대감입식시각특정진행운동속성분석화도상처리,실현인원감공특정분석산법개진。방진결과표명,해계통능유효실현원거리시빈감공화인원특정식별급감입식시각정위,측거오차교저,대군사금구인원적감공정도제고,대원정감공시빈특정적장경회복성능교호,능유효식별인원적행위화시각특정,성능우월。
To improve the personnel monitoring and recognition ability of the military restricted zones through the visual characteristics of the military restricted zone personnel extraction. Extraction of traditional system design and personnel vi?sual feature algorithm uses video step tracking study method, pixel feature extraction and classification of personnel by SVM lock personnel location information, due to the randomness and uncertainty of the staff, resulting in visual recognition effect is not good. the military restricted zone personnel monitoring technology of the embedded vision is proposed based technology system using 4 computers, 3 monitors, virtual reality equipment design, realization of improved analysis algo?rithms personnel monitoring feature. Simulation results show that the system can effectively realize remote video monitoring and personnel feature recognition and embedded visual positioning, ranging error is low, improve the monitoring accuracy of the military restricted zones of personnel, on the remote video monitoring characteristics of the scene recovery perfor?mance is good, can the behavior and visual features effectively identify people, it has superior performance.