The method of automatic clarity detection for aerial remote sensing based on the spatial-filtering velocimetry can effectively solve the problem of inability to focus on static targets during aerial camera flights. This method can flexibly select the type of transmission function and filter window configuration, and dynamically adjust the filter parameters. To address the issue of selecting the configuration of a window for detecting the clarity of aerial remote sensing based on spatial-filtering velocimetry, this study constructs typical spatial filter models from a mathematical perspective and analyzes the sinusoidal and rectangular transmission functions. Sinusoidal and rectangular transmission functions have similar low-pass filtering characteristics, but the rectangular transmission function contains high-frequency components in addition to the fundamental peak. For clarity detection, ground targets satisfy a random process, and the high-frequency components they contain are beneficial for improving detection accuracy. Therefore, theoretically, the use of a rectangular-type filter is more conducive to detection accuracy, and the parameters of the rectangular transmission function are easier to design, with a smaller computational load, making it more suitable for automatic clarity detection. The study analyzes the power spectral density functions and filtering characteristics of typical spatial filter configurations, including rectangular, circular, and Gaussian-weighted filters, and discusses the impact of different window forms of filters on the accuracy of aerial remote sensing clarity detection. From the perspective of window configuration, rectangular and circular spatial filters contain low-frequency components that are unfavorable for clarity detection, as well as high-frequency components that are beneficial for the detection results. The Gaussian-weighted spatial filter does not contain these low-frequency components, but the low-frequency components can be filtered out using a differential method, and the energy of the high-frequency components is small, which will not have a significant impact on the detection results. Therefore, there is no significant difference in the performance of these three typical window configurations when used for aerial remote sensing clarity detection. In terms of implementation difficulty, the rectangular-type spatial filter is easier to realize, with a smaller computational data volume and more convenient parameter adjustment, and is therefore more commonly used. A validation experiment was designed by fixing the camera lens on a precision guide rail and using a high-precision turntable to drive its rotation, simulating the dynamic imaging process of aerial remote sensing. The imaging characteristics of an area array CCD detector were utilized, and dynamic sampling was performed along the periodic transmittance direction of the filter to simulate the periodic modulation process of the light amplitude of moving images relative to the target speed, and interval sampling of the CCD image was performed to simulate the periodic transmittance ratio of the spatial filter, greatly simplifying the system structure compared to using physical spatial filters such as gratings. The experiment first manually calibrated the focal plane as the zero position, using the rectangular transmission function and the common spatial filter configurations, including rectangular window, circular window, and Gaussian-weighted window, to perform ground imaging automatic clarity detection experiments. For each window type, 11 positions from out-of-focus to in-focus and back to out-of-focus were captured, and evaluation curves were generated, with the peak of the curve corresponding to the best imaging clarity position of the camera. The best imaging clarity positions obtained from the experimental curves of the three filters were consistent with the manually calibrated focal plane position, and the step size between the two frames in the experiment was 60 μm, less than the optical system's half-depth of focus of 76.8 μm. The experimental results show that the clarity detection results using the three filter configurations are consistent, with single-peak and unbiased characteristics, and no significant performance differences, so the rectangular-type spatial filter, which is the easiest to implement and requires the least computational resources, can be selected.