Research of Application on Deblurring model for Guqin Performance

In order to improve the quality of fingering fuzzy image enhancement in guqin performance and further promote the digital research work on guqin art, a two-stage method of hand analysis network and SNAD generator is proposed to solve the fingering fuzzy problem in guqin performance. First of all, aiming at the feature that the camera does not move and the finger moves, resulting in finger motion blur, a hand analysis mesh is proposed to be constructed. Second, an adaptive denormalization network is constructed to achieve accurate reconstruction of the hand structure. Finally, an adaptive denormalization network is constructed to achieve accurate hand structure reconstruction. Experimental results show that the performance of our non-uniform blur model on real data can be further improved, similar to traditional iterative optimization-based BID methods, but slower than emerging instance-driven methods.