When detecting MTF from satellite remote sensing image, it is time-consuming to select e edge of natural objects manually, and easy to be disturbed by human factors, so the test results are unstable. Aimed at eliminating above problems, an automatic MTF detection method for satellite remote sensing images based on Non-Subsampled Shearlet Transform (NSST) is proposed in this paper. Firstly, the image is decomposed into multi-scale and multi-directional components, and multi-scale product filtering and composite morphological operators are used to process the high and low frequency component images respectively. Then conduct research on the entire process from edge image extraction to MTF calculation, gradually screening the edges. Finally, the NSST edge extraction method isused to compare the MTF detection results of medium and high-resolution images with various typical edge operators. The method is applied to target, medium, and high-resolution images for MTF detection. The experimental results show that the NSST edge extraction method removed invalid information such as gradient areas and noise, and the obtain edge lines meet the detection requirements of clear edge lines, uniform edges, and a certain range of inclination angles in the edge method. The MTF automatic detection results applied to target images are consistent with the published data. The standard deviation of MTF automatic detection results for the same terrain edge area is 0.0043 multiple times, and the standard deviation is manually calculated using general software to be 0.0101. The satellite remote sensing image MTF automatic detection method based on NSST has stronger error stability and is suitable for target images. Through multi scenario calculation and statistical analysis, the MTF detection efficiency has been enhanced more than double.