The benchmark of image denoising model works indicates significant improvements in noise removal as evidenced by PSNR and SSIM scores attained across the three major noise categories of Gaussian, Poisson, and Salt-and-Pepper types of noises. For Dataset 1, the model achieved PSNR of 32.45 dB and the SSIM of 0.910 while current benchmarks for similar models with similar reliance on deep learning techniques had PSNR 31.12 dB and average SSIM 0.890. The proposed method was tested on Dataset 2 yielding a PSNR of 30.98 dB and SSIM of 0.875 which is superior to current approaches. The ablative analysis reveals that the main components including HRD framework and AENS mechanism are both essentially important because their removal degrades the PSNR by up to 1.7dB. Another factor of economy which has been enhanced at 30 ms is the time taken to make an inference based on the model. These results together corroborate the model’s ability as well as speed to perform denoising tasks and provides a robust ground for further advancements in the field of image processing, particularly for those applications where noise reduction is an important and time sensitive operation.