分析了基于特征点控制为基础的配准、傅里叶梅林变换、最大互信息法等三种图像配准算法的原理。采用这三种方法对近红外脉冲光乳腺成像系统采集的病例图像实现图像配准,并从配准所需要时间和最终的配准效果两个方面对三种算法的配准效果进行了比较。结果表明,基于特征点控制基础的配准方法所需时间最短,傅里叶梅林变换次之,最大互信息法所需要的时间最长。但是,从最终配准效果来看,最大互信息法明显优于其他两种算法。应用最大互信息法可以有效实现图像的配准,但是在实际应用中,需要对配准时间进行进一步优化。 The principle of three kinds of image registration algorithms based on feature point control, Fourier Mellin transform and maximum mutual information method is analyzed. The three methods were used to realize the image registration of case images collected by near infrared pulsed light breast imaging system. The registration results of the three algorithms were compared from the two aspects of registration time and final registration effect. The results show that the registration method based on feature point control takes the shortest time, followed by Fourier Merlin transform, and the maximum mutual information method takes the longest time. However, the maximum mutual information method is obviously superior to the other two algorithms from the final registration effect. The application of the maximum mutual information method can effectively achieve image registration, but in practical applications, the registration time needs to be further optimized.