CAJ | 학술논문

研究了基于视频图像的目标识别与跟踪算法。主要应用了Hog特征和支持向量机（SVM）分类器设计完成对目标的侦察识别。当侦察到目标时，采用高斯模型对目标和背景建模，结合卡尔曼滤波预测估计方法和均值漂移（mean-shift）理论，实现目标跟踪和定位功能。最后通过缩微环境中的案例演示，验证了所提算法的可行性。
연구료기우시빈도상적목표식별여근종산법。주요응용료Hog특정화지지향량궤（SVM）분류기설계완성대목표적정찰식별。당정찰도목표시，채용고사모형대목표화배경건모，결합잡이만려파예측고계방법화균치표이（mean-shift）이론，실현목표근종화정위공능。최후통과축미배경중적안례연시，험증료소제산법적가행성。
This paper investigates the algorithms of target recognition and target tracking based on video sequence. We mainly focuses on the Histogram of Orientation Gradient (HOG)Algorithm and the Support Vector Machine (SVM)Algorithm for learning and classifying to implement the target recogmition. When the target is recognized,using Gauss models of target and background,combining Kalman filter used to predict tartet position and mean-shift theory,we completes the target tracking function. At last,the experiments in down-scaled environment proves the correctness of the method.