In order to improve the stability of the characteristic spectral lines of laser-induced breakdown spectra of heavy metals in soil and the accuracy of soil quantitative analysis, the image optimization technology and laser-induced plasma technology were combined to analyze element Cu in soil. The two characteristics lines, Cu Ⅰ 324.75 nm and Cu Ⅰ 327.40 nm, were compared and analyzed by experiments. Finally, Cu Ⅰ 324.75 nm was selected as the analytical spectral line. The method of wavelet transform was used to denoise the spectrum, which eliminates the influence of matrix effect on the results and improves the stability of the spectral. Then the real-time acquisition of plasma images with different delays was carried out. The influence of delays on spot area and spectral intensity was analyzed, and the optimal delay was determined to be 900 ns. The RSD of different concentrations with the image optimization model was compared with that without optimization at the optimal delay time and the same energy. The optimal plasma image was selected by the image optimization model to calculate the RSD. It was found that RSD with improved greatly. The RSD of each concentration without optimization was 5.39%, 6.22%, 7.56%, 8.42% and 9.63%, respectively. The RSD with optimization was 3.24%, 4.47%, 5.32%, 6.13% and 7.21%, respectively. The method of image optimization effectively suppresses the continuous background radiation and improves the stability and repeatability of the spectrum. Compared with the data without image optimization, the RSD with optimization decreased by 2.15%, 1.75%, 2.24%, 2.29% and 2.42%, respectively. The date proved that the stability of detecting the content of copper in soil was greatly improved. Finally, the standard internal method was used to analyze the heavy metals in soil quantitatvely. Compared with the non-optimal condition, the accuracy and stability of the calibration model under the optimal condition were improved. The R² increased from 0.978 to 0.995. It can be seen from the data above that image optimization technology greatly improves the spectral stability. The image optimization technology can greatly improve the quantitative analysis ability in soil heavy metal detection using LIBS.