Recent advancements in image dehazing, especially within indoor environments, are highlighted by the SOTS-Indoor dataset. Traditional methods, such as MSBDN, FFA-Net, DeHamer, and MAXIM-2S, have employed techniques like feature fusion, attention mechanisms, and multi-scale feature extraction to address the challenge of haze removal. MSBDN attained a Peak Signal-to-Noise Ratio (PSNR) of 33.67 and a Structural Similarity Index (SSIM) of 0.985 on the SOTS-Indoor dataset, while FFA-Net improved these metrics to 36.39 and 0.989, respectively. Subsequent models, including DeHamer and MAXIM-2S, continued to enhance performance, achieving PSNR values of 36.63 and 38.11. This study introduces DehazeCLNet, a novel model integrating contrastive learning to enhance haze suppression capabilities. By incorporating contrastive loss and feature extraction across multiple depths, DehazeCLNet achieves notable image restoration, with a PSNR of 42.57 and SSIM of 0.996, surpassing previous methods. These results underscore DehazeCLNet’s potential, establishing new benchmarks for indoor image dehazing and suggesting promising directions for future research in haze removal.