结合PCA的尺度不变特征变换(SIFT)算法，详细画出了时域和频域的相关图，车牌识别定位程序的部分功能，最大似然（ML）准则和最大后验概率（MAP）准则，可实现对二维数据的聚类，本程序的性能已经超过其他算法。- Combined with PCA scale invariant feature transform (SIFT) algorithm, Correlation diagram shown in detail the time domain and frequency domain, Part of the license plate recognition locator feature, Maximum Likelihood (ML) criteria and maximum a posteriori (MAP) criterion, Can realize the two-dimensional data clustering, This program has exceeded the performance of other algorithms.