The environment of flotation site is harsh， and the lighting conditions are complicated and variable. For the flotation images collected on-site were prone to underexposure， color distortion， and other problems， a low-illumination image color depth coding and decoding correction and multiscale enhancement method was proposed. First， the low-illumination image was converted from RGB to HSV space， and for the value （V） component， the non-subsampled shearlet transform （NSST） was used to perform multi-scale decomposition. Second， a color depth codec correction network based on the global spatial module was proposed， and the color depth coding and decoding correction network model was constructed through squeeze extraction， color coding， color decoding， and color correction. Color correction was performed for the hue （H） and saturation （S） components. Then， the low-frequency subband images of the V component were enhanced by adaptive fuzzy set， and the noise components of each high-frequency subband image of the V component were effectively filtered out using the scale correlation coefficient， while the high-frequency edge coefficients were significantly enhanced using the nonlinear gain function. Finally， the enhanced images of the various subbands of the V component were reconstructed by the inverse transform of the NSST， and the reconstructed V component was fused with the corrected H and S components and converted back to RGB space. Experimental results demonstrate that， compared with current mainstream methods， the proposed method achieves an average reduction of 14.835 8 in CIEDE， an average increase of 8.48 dB in PSNR， and an average improvement of 31.32% in structural similarity， with the continuous edge pixel ratio maintained above 91%. The proposed method enhances image brightness， achieving higher contrast， clarity， and information entropy. The image colors are closer to true colors， preserving more texture details， and edges are enhanced while effectively suppressing noise.