Research on Image Dehazing Algorithm Based on Fusion Transformer Block

Abstract

Given the high standards of image quality demanded today, there is still room for improvement in the detail processing of image dehazing technology. An improved image dehazing model based on Transformer architecture is proposed in this paper. The model's performance is enhanced through an augmented attention mechanism and the integration of residual blocks. The shortcomings of existing dehazing algorithms are initially analyzed, whereupon the design principles and detailed implementation of the enhanced model are introduced. Subsequently, the effectiveness of the model in performing image dehazing tasks is experimentally validated, showcasing its efficacy in enhancing image quality and preserving intricate details.