【摘要】 针对帧差法和背景差分法检测运动目标准确率低,自适应能力弱等缺陷,提出了一种改进五帧差分法与背景差分法和模板匹配相结合的运动目标检测和识别算法;通过改进的五帧差分和背景差分法融合的算法从视频图像序列中检测出运动目标;利用形态学方法去除噪声,改善运动目标提取效果;在Harris算法提取图像匹配特征值的基础上角点配准,提高图像识别的准确率,通过提取目标特征与自适应模板图像进行特征匹配的方法实现了目标检测识别和跟踪。仿真结果和实验表明该方法有噪声和部分遮挡的运动目标有良好的检测识别效果,识别率达到了95%。更多还原

【Abstract】 Aiming at the disadvantages of frame difference method and background difference method for detecting moving objects with low accuracy and weak adaptive ability,a moving object detection and recognition algorithm based on improved five-frame difference method combined with background difference method and template matching was proposed. The moving object is detected from the video sequence by combining the improved five frame difference and background difference algorithm. Using morphological method to remove noise and improve the effect of moving target extraction. On the basis of Harris algorithm extraction of image matching eigenvalues,corner registration is used to improve the accuracy of image recognition. The target detection recognition and tracking is realized by the method of matching the feature of the target and the adaptive template image. The experiment shows that the method is simple and effective,and has a good detection and recognition effect for moving targets with noise and partially shielded,the recognition rate reaches 95%.更多还原