Abstract:The surge in interest regarding image dehazing has led to notable advancements in deep learning-based single image dehazing approaches, exhibiting impressive performance in recent studies. Despite these strides, many existing methods fall short in meeting the efficiency demands of practical applications. In this paper, we introduce WaveDH, a novel and compact ConvNet designed to address this efficiency gap in image dehazing. Our WaveDH leverages wavelet sub-bands for guided up-and-downsampling and frequency-aware feature refinement. The key idea lies in utilizing wavelet decomposition to extract low-and-high frequency components from feature levels, allowing for faster processing while upholding high-quality reconstruction. The downsampling block employs a novel squeeze-and-attention scheme to optimize the feature downsampling process in a structurally compact manner through wavelet domain learning, preserving discriminative features while discarding noise components. In our upsampling block, we introduce a dual-upsample and fusion mechanism to enhance high-frequency component awareness, aiding in the reconstruction of high-frequency details. Departing from conventional dehazing methods that treat low-and-high frequency components equally, our feature refinement block strategically processes features with a frequency-aware approach. By employing a coarse-to-fine methodology, it not only refines the details at frequency levels but also significantly optimizes computational costs. The refinement is performed in a maximum 8x downsampled feature space, striking a favorable efficiency-vs-accuracy trade-off. Extensive experiments demonstrate that our method, WaveDH, outperforms many state-of-the-art methods on several image dehazing benchmarks with significantly reduced computational costs. Our code is available at https://github.com/AwesomeHwang/WaveDH.
Abstract:Recent work on super-resolution show that a very deep convolutional neural networks (CNN) have obtained remarkable performance. However, as CNN models have become deeper and wider, the required computational cost is substantially higher. In this paper, we propose Linear Depthwise Convolution to address this problem in single image super resolution. Specifically, Linear Depthwise Convolution can reduce computational burden on CNN model, preserving information used to reconstruct super-resolved image. The performance improvement of our proposed method is due to removing non-linearity between depthwise convolution and pointwise convolution. We evaluate the proposed approach using Set 5 and Set 14 datasets and show it performs significant better performance.
Abstract:Thermal cameras shows noisy images due to their limited thermal resolution, especially for scenes of low temperature difference. In this paper, to deal with noise problem, we propose a novel neural network architecture with repeatable denoising inception residual blocks(DnIRB) for noise learning. Each DnIRB has two sub-blocks with difference receptive fields and one shortcut connection for preventing vanishing gradient problem. The proposed approach is tested for thermal images. The experimental results show that the proposed approach show the best SQNR performance and reasonable processing time compared with state-of-the-art denoising methods.