Abstract:With the increasing scale and complexity of modern distribution networks, security monitoring is facing serious challenges, especially data management and redundancy. In this paper, an OpenPose -GLCM-hash data cleaning technology is proposed, which combines the global texture features of GLCM and the human pose features of improved OpenPose algorithm, and generates an image hash code through the weighted combination of these two features to achieve image de-weighting. The experimental results show that OpenPose -GLCM-hash algorithm has high weight removal rate and low error rate in different environments, especially in noisy environments. This research provides an efficient solution for data management of power distribution safety supervision.