Monitoring of extrusion filament state for fused filament fabrication: A super-resolution image monitoring approach based on degradation pattern learning

Extrusion filament is the fundamental unit of the fused filament fabrication process (FFF). Its state stability directly affects the product quality. Therefore, a reliable method is required to monitor the state of the extrusion filament. This work presents an innovative approach involving super-resolution algorithms for conveniently monitoring the state of the extrusion filament during the printing process using cost-effective cameras. Firstly, this work captures high-resolution images during the printing ending and low-resolution images during the printing processing. Secondly, a generative adversarial network (GAN) is employed to adaptively learn the degradation patterns in the printing environment. Subsequently, the trained generator of the GAN is used to construct pixel-level paired super-resolution training dataset. The work then utilizes this dataset to train a super-resolution model, getting a specialized super-resolution model for the degradation patterns specific to the printing scenario. Finally, this super-resolution model is employed to enhance the resolution of the images captured during the production process, resulting in improved performance in downstream image-related tasks.