A denoising neural network based on wavelet transform (DWTNet) was used to study the denoising of core images in a sand and mud interlayer core of a certain block in the Ordos Basin. The evaluation results of this method were compared using peak signal-to-noise ratio (PSNR) and denoised image results. Research has shown that the proposed algorithm using wavelet transform denoising neural network (DWTNet) was tested on test sets YX1 and YX2, and compared with denoising algorithms such as EGDNet. When the noise levels were 25, 50, and 75, PSNR was 0.527 dB , 0.418 dB , and 1.1 dB higher than the EGDNet algorithm. The proposed algorithm outperforms other algorithms in terms of peak signal-to-noise ratio and other indicators; And in terms of visual effects, the processed images are also clearer. The proposal of methods is of great significance for porosity, average volume specific surface area, and average curvature calculation.Keywords: digital core technology, Sandstone, Wavelet transform, Neural network, Denoising; Rock characteristi