Peer-to-peer networking is a scalable method of distributing multimedia files via the internet. Content-based retrieval in peer-to-peer networks is a critical but difficult topic when there is a big amount of visual data spread among multiple nodes. While most previous methods rely on indexing high-dimensional visual features and have scalability restrictions, the bag-of-visual-words model is used in this paper to offer a scalable strategy for content-based picture retrieval in peer-to-peer networks. Because images are spread over the whole peer-to-peer network, obtaining a global codebook is more difficult than in centralized setups. Furthermore, because a peer-to-peer network frequently evolves dynamically, a static codebook is less useful for retrieval tasks. As a result, we offer a method for dynamic codebook updating that optimizes mutual information between the final codebook and relevance information, as well as workload balance across nodes that manage distinct code words. Indexing trimming strategies are being developed in order to increase retrieval performance and lower network costs. The proposed approach is scalable in developing and distributed peer-to-peer networks, with enhanced retrieval accuracy, according to our extensive experimental data.