推扫式高光谱相机自动调焦方法(中国光学工程学会推荐稿件)

自动调焦方法已广泛得应用到各种光学仪器中。为了解决推扫式高光谱相机自动调焦问题，提出了一种利用基于四元数小波变换的光谱质量评价函数估计相机焦距的方法。该方法首先将推扫线上单行光谱数据顺序排列成一个二维矩阵，利用四元数小波变换对光谱像元矩阵进行分解得到低频和高频四个子带。通过对低频和高频的幅值和相位信息构造光谱质量评价函数，指导相机自动调焦。调焦过程中，利用自动调焦机构按一定步长旋转镜头，计算不同离焦状态下的光谱质量指标值，拟合光谱质量指标值与镜头伸缩量之间的高斯分布函数，最终估算最佳调焦量。实验表明，提出的光谱质量评价函数灵敏度高、准确度强。该调焦方法仅利用单行光谱数据进行调焦，调焦准确。

Auto-focusing method of push-broom hyperspectral camera

Auto-focusing method has been widely used in various optical instruments. Aiming at the problem of auto-focusing of push-broom hyperspectral camera, a method is proposed to estimate the focus of a camera by using the spectral quality evaluation function based on quaternion wavelet transform. Firstly, the single-line spectral data of the push sweep line is arranged into a two-dimensional matrix, and the matrix is decomposed into four low-frequency and high-frequency subbands. The spectral quality evaluation function is constructed by using the low-frequency and high-frequency amplitude and phase information to realize auto-focus. In the process of focusing, the spectral quality metrics is calculated in different focuses by rotating the lens at a certain stride length by an auto-focus mechanism. A Gaussian distribution function is fitted between the spectral quality metrics and the amount of expansion and contraction of lens. Finally, the optimal focus is estimated. Our results have shown that the proposed evaluation function of spectral quality has better sensitivity, accuracy. The auto-focus method has good accuracy due to using only a single line of spectral data for focusing.