EvTexture: Event-driven Texture Enhancement for Video Super-Resolution

Event-based vision has drawn increasing attention due to its uniquecharacteristics, such as high temporal resolution and high dynamic range. Ithas been used in video super-resolution (VSR) recently to enhance the flowestimation and temporal alignment. Rather than for motion learning, we proposein this paper the first VSR method that utilizes event signals for textureenhancement. Our method, called EvTexture, leverages high-frequency details ofevents to better recover texture regions in VSR. In our EvTexture, a newtexture enhancement branch is presented. We further introduce an iterativetexture enhancement module to progressively explore thehigh-temporal-resolution event information for texture restoration. This allowsfor gradual refinement of texture regions across multiple iterations, leadingto more accurate and rich high-resolution details. Experimental results showthat our EvTexture achieves state-of-the-art performance on four datasets. Forthe Vid4 dataset with rich textures, our method can get up to 4.67dB gaincompared with recent event-based methods. Code:https://github.com/DachunKai/EvTexture.