摘要
提出了一种基于小波变换和主成分分析(PCA)方法实现的车辆静态图像检测方法。利用选定的车辆和背景训练样本集,用小波变换对样本进行特征提取,通过主成分分析来设计适当的分类器,把待检测图像在多分辨率下逐块进行分类,以此来判断某区域内是否有车辆,完成车辆检测的任务。实验结果表明,这种方法实现简单、应用效果良好,具有较好的应用前景。
This paper presents an algorithm for detecting cars from static road images based on statistical pattern recognition. Wavelet transform is employed to extract features from the original images with supervised principal component analysis (PCA) used to design the classifier. The principal component spaces of the car class and the background class are computed separately. The classes are identified according to the residues of a sample projected onto these two spaces. The original image is divided into many subregions with different resolutions which are then scanned by the classifier. Experimental results with real images show that the performance is quite good.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2002年第11期1560-1564,共5页
Journal of Tsinghua University(Science and Technology)