中文摘要：

车牌照识别技术是现代智能交通的重要组成部分．为了从彩色图像中定位及分割出汽车牌照，将图像中每一个像素点的红、绿、蓝值看成三维空间中的点（r，9，6），采用主成分分析对含牌照的彩色图像进行正交变换得到（r’，g’，b／），分别以r’，g’，b’作为灰度值得到三个主成分图像，牌照信息体现在第二主成分图像中．为了进一步提高车牌定位效率，对正交变换矩阵进行了修正，对修正后的第二主成分图像进行二值化，分割出白色区域即可定位出汽车牌照．实验表明，本算法对噪声不敏感，抗干扰性强，识别效率高．

英文摘要：

License plate location plays an important role in recognition of modern vehicle license plates. In order to locate and segment the license plate from the colored images, we firstly consider the red, green and blue value of every pixel in images as the point of the three-dimensional space （r, g, b）, respectively. Secondly, （r＇, g＇, b＇） is derived from orthogonal transformation of the colored images including license plate by means of principal component analysis, where r＇, g＇, b＇ represent the gray levels of three principal component images, respectively and the information of the license plate is indicated in the second principle component. With improving the efficiency of license plate location, we modify the orthogonal transformation. Finally, we can locate the vehicle plate through segmenting the white area of the binary image of the modified second principle component. Numerical experiments illus- trate that the algorithm is not sensitive to noise, strongly anti-jamming and high efficiency of recognition.