基于四元数小波变换的清晰度评价

Quaternion wavelet transform for evaluation of image sharpness

针对当前各种图像清晰度评价方法在清晰度判别过程中单调性和区分度不够以及适用范围较小的问题,提出了一种基于四元数小波变换(QWT)幅值与相位的图像清晰度评价方法。该算法通过四元数小波变换将图像从空间域变换到频率域,对得到的四元数小波变换系数进一步计算之后获得低频子带与高频子带的幅值与相位信息,求得低频子带幅值各方向的梯度之后与对应方向的相位相乘求和,最终得到两个清晰度指标值。采用该算法与多种现有算法对不同内容的图像、不同程度模糊的图像以及含有不同程度噪声的图像进行清晰度评价实验:相对于现有算法,所提算法在对上述多种图像的清晰度评价中都保持着很好的单调性与区分度。实验结果表明,所提算法不但克服了现有算法在单调性与区分度上的不足,而且所提清晰度评价指标可以应用在图像处理中。

The current image sharpness evaluation methods suffer from insufficient discrimination and monotonicity as well as the narrow applicable scope in the process of measurement. In order to solve the problems, an image sharpness evaluation method based on amplitude and phase of Quaternion Wavelet Transform（ QWT） was proposed. The image was transformed into frequency domain from spatial domain by QWT, the amplitude and phase information of low frequency subband and high frequency subbands were got by further calculation of the QWT coefficients. After multiplication of the gradient of amplitude and corresponding directional phase, the proposed sharpness metrics were obtained by adding each direction＇s value together. The proposed sharpness metrics were conducted on the images with different content, degree of blur and degree of noise. Compared with the existing algorithms, the proposed algorithm has a fine monotonicity and discrimination for all kinds of images. The experimental results show that the proposed sharpness metrics not only overcome the shortcomings of the existing algorithms in the monotonicity and discrimination, but also can be applied to image processing.