In the design and research of an unmanned aerial vehicle-borne hyperspectral imager with an acousto-optic tunable filter (AOTF), the depth of the field range of the imaging objective lens is fixed, the spectral range of 400-1000nm is wide, and the AOTF crystal has different deflection angles for diffracted light of different wavelengths. These features all cause problems such as image blurring and spectral drifts. Therefore, we propose a solution using a tunable zoom lens and multi-modal spectrum registration to solve these problems. We also propose an improved image definition evaluation function using the zoom lens to obtain clear horizontal lines and blurred vertical lines and, thereby, hope to overcome image blurring in an AOTF spectrometer. The zoom parameters corresponding to the clear image in each spectral segment of the AOTF spectrometer can be determined rapidly and accurately. However, it is difficult to establish a mathematical model of the drift, using wavelength as a variable, to compensate the drift due to the clear image of each spectrum obtained using the zoom lens. The calibration method completes the registration of the spectral data cube composed of clear images, solving the problem of spectral band drift. Moreover, it has been verified through UAV-borne hyperspectral imaging that image blurring and spectral drift of the AOTF imaging spectrometer were significantly reduced, providing opportunities for the application and research of AOTF-based hyperspectral imagers in more fields, such as remote sensing.