Volume 5, Issue 1 (2025) – 5 articles
Cover Picture: This paper proposes a novel non-uniform image dehazing algorithm based on serialized integrated attention and a multi-dimensional Transformer to address detail loss and blurred restoration issues in existing methods. The approach integrates spatial and channel attention in a shallow-layer network to capture local features while employing a multi-dimensional Transformer in a deep-layer network to extract global information. By adaptively fusing features from both layers, the model enhances fine-grained details and overall image context. Experimental results on multiple haze datasets demonstrate superior performance in both visual quality and objective metrics, outperforming existing dehazing algorithms.
view this paper 
