<p>Backpropagation neural networks (BPNNs) can be used to restore images; however, the error surface of the BPNN algorithm contains several local minima, making it easy to slip into suboptimal solutions. A genetic algorithm (GA) with a strong global searchability can optimize the initial weight and threshold of BPNNs. However, traditional GAs are prone to local convergence and stagnation; hence, this article proposes a hybrid GA. First, the hybrid GA introduces an elite opposition-based learning strategy to enhance population diversity and prevent premature convergence. Second, the firefly algorithm updates mutated individuals twice. Thus, the searchability of the algorithm in the vicinity of the optimal solution is increased. The proposed IGABPR approach outperforms both BPR and GABPR in terms of peak signal-to-noise ratio (PSNR) and mean squared error (MSE) across ten test photos. For example, it provides better PSNR values (e.g., 36.73 vs. 36.59) and lower MSE values (e.g., 13.80 vs. 14.26), implying more accurate and dependable restoration. These findings demonstrate the success of hybrid GA optimization to enhance prediction accuracy and Adaptability in image reconstruction tasks. The results show that the BPNN-based image restoration algorithm optimized with the improved genetic algorithm outperforms both the standard BPNN-based algorithm and the version optimized with the traditional genetic algorithm in terms of peak signal-to-noise ratio and mean squared error.</p>