论文部分内容阅读
直接使用傅里叶梅林变换计算出的图像间旋转参数的精度不高,以此旋转参数对图像进行纠正后再使用相位相关的扩展方法计算出的图像间的平移参数精度也不高.针对这种情况,提出一种改进的角度检测方法.传统的方法是先对图像进行傅里叶变换,然后进行极坐标变换,再使用相位相关得到相关谱,通过相关谱的峰值位置确定图像间的旋转角度,最后对图像做角度纠正并计算平移参数.而本文对相关谱峰值所在区域进行拟合,使用拟合后的峰值位置替代原来的峰值位置确定图像间的旋转角度.由于拟合后的峰值位置的精度高于拟合前,估计角度的精度会更高,使用此角度做旋转纠正能够提高图像之间的相关性,进而也提高平移参数的精度.实验证明:与Fourier-Mellin变换的通常做法相比,此方法显著提高了配准的精度,平均角度精度提高0.187°,水平方向上平均平移精度提高0.339 pixel,垂直方向上平均平移精度提高0.296 pixel.
The accuracy of the rotation parameters calculated by the direct Fourier transform is not high, so the accuracy of the translation parameters between the images calculated by using the phase-dependent expansion method after the rotation parameters correct the image is not high. We propose an improved method of angle detection.The traditional method is to first Fourier transform the image and then transform the polar coordinate and then use the phase correlation to obtain the correlation spectrum and determine the rotation between the images by the peak position of the correlation spectrum Angle, and finally the angle of the image is corrected and the translation parameter is calculated. In this paper, the area where the correlation peak is located is fitted and the rotated position is replaced by the peak position after fitting. Since the peak value after fitting The accuracy of the position is higher than that before fitting, and the accuracy of the estimated angle will be higher. Using this angle to do the rotation correction can improve the correlation between the images and improve the accuracy of the translational parameters. Experimental results show that, compared with the usual Fourier-Mellin transform, Compared with the method, this method significantly improves the registration accuracy, the average angle accuracy is improved by 0.187 °, and the average horizontal translation accuracy is improved 0.339 pixel, the average vertical translation accuracy of 0.296 pixel.