Research on the dynamicization of Qijiang block printing based on image segmentation and pose estimation

Qijiang prints record the folk customs of Bayu with the technology of "multi-color printing in one edition". Its rough shape and bright colors bear the cultural memory of farming civilization, but the static display is difficult to inspire young audiences to resonate. In this study, the technical path of "single subject dynamic" is put forward. Through image segmentation and posture estimation technology, the main characters in printmaking are transformed into animated characters that can synchronize their actions in real time, while the static characteristics of background and secondary characters are preserved, so as to demonstrate the necessity and feasibility of technical intervention in activating the intangible heritage. Aiming at the characteristics of high saturation color block and exaggerated modeling of printmaking, the U-Net model is improved and combined with color threshold segmentation to extract the main outline accurately and retain the decoration of powder printing. The lightweight attitude estimation framework cooperates with the skeletal animation system to realize the real-time synchronization of audience movements and printmaking characters. This method achieves 85% segmentation accuracy on Qijiang print data set, and the fusion of dynamic subject and static background avoids visual separation. Through the design of "local dynamic+overall static", this strategy not only protects the integrity of the original, but also makes the audience intuitively perceive the narrative tension of folk scenes through action imitation. Single-agent dynamic is not only a technical experiment, but also a non-genetic innovation paradigm-to achieve maximum benefit with minimum intervention and provide a solution with both cultural fidelity and communication effectiveness for the digital transformation of traditional art.