In this paper, we introduce a method for denoising high-definition images using a block-matching 3D double convolutional neural network (BM3D-C2L-Net). Depending on the weather, HDR photos may exhibit a wide variety of dynamic noise and rain streak patterns. Many researchers have found it difficult to find a solution for removing high density, directed noise patterns from photographs. In order to solve this issue, a convolutional neural network structure with an extra "block" is devised to make sure that inputs from the previous layer are used in the next. To improve the improvement of the image despite the noise, a dual convolution layer structure is adopted. We test the BM3D-C2L network's denoising capabilities on a range of image sequences with varying levels of noise and rain streak patterns. The double convolutional layer approach makes it simple to train a network to remove the directional oriented noise pattern. The suggested denoising network is trained on the TensorFlow open-source platform. According to the experiments, the suggested draining network provides a greater PSNR value than any other existing denoising approach