【摘要】 采用光学相关、人工神经网络算法和空域执行的综合鉴别滤波器相结合的光电混合图象处理技术 ,对三维运动目标 (模型 )进行实时识别与跟踪 .实验结果显示 :在强背景噪声干扰下 ,目标的识别速率最大可达到 16次 /s;定位精度优于± 1个象素 ;正确识别率超过 95 % .实现了在强背景噪声干扰下三维运动目标的实时畸变不变高可靠性的自动识别与精确定位 ,解决了纯电子模式识别技术难以解决的问题更多还原

【Abstract】 The digital-optical hybrid image processing technique has many advantages compared with the electronic processing,such as high operation speed,high resolution,low power consumption,high recognition efficiency,low false alarm and good clutter rejection ability etc. It could be used to perform distortion-invariant recognition and tracking of moving-object in complex noisy background automatically.The optical correlation,artificial neural network (ANN) and synthetic discriminant filter(SDF) employed in spatial domain are used to make recognition and tracking of the 3-D moving object models.The experiment results show that even in strong noise background,this schem has excellent recognition and pointing performances with max recognition speed: 16 frames/s,location accuracy: Δx<±1,Δy<±1; and recognition probability >95%. The real-time distortion-invariant recognition and tracking to the moving-object in noisy background are realized with high reliability,which is very difficult for electronic processing.更多还原