基于定向梯度直方图及支持向量机识别篮球裁判信号的研究

Research on Basketball Referee Signal Recognition Based on Directional Gradient Histogram and Support Vector Machine

实时系统中用于人机交互的手势,无论是静态的还是动态的,都是一个活跃的研究领域,并具有许多可能的应用.应用于实时的基于视觉的手势接口需要快速且极其健壮的手势检测和识别.然而,这种手势检测和识别对图像分割技术提出了巨大的计算要求.采用支持向量机(SVM)进行分类的一种基于方向梯度直方图(HOG)特征的图像分割技术,可以从视频中识别出篮球裁判的信号,分类识别正确率可达97.5%.

Gestures for human-computer interaction in real-time systems,whether static or dynamic,are an active research field and have many possible applications.However,vision-based gesture interfaces for real-time applications require fast and extremely robust gesture detection and gesture recognition.Trying to recognize the gestures of officials in typical sports videos puts forward huge computational requirements for image segmentation technology.In this paper,an image segmentation technique based on directional gradient histogram(HOG)feature is proposed,which can recognize the signal of basketball referee from video.Support vector machine(SVM)is used for classification,and the classification accuracy is 97.5%.