聚类连接图的构建方法、装置、设备及可读存储介质
By performing analysis of variance in image processing to determine the similarity threshold and constructing a clustering connectivity graph, the problem of connectivity graph error in existing technologies is solved, and higher quality image clustering is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- QINGDAO INTELLIFUSION TECH CO LTD
- Filing Date
- 2022-06-14
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, it is impossible to set an accurate similarity threshold for each category of images when constructing a connection graph, which leads to errors in the connection graph. This results in connection edges between image nodes that do not belong to the same category, while connection edges are missing between image nodes that belong to the same category.
By obtaining similarity values from multiple frames of images, performing variance analysis, determining the maximum test bias to obtain a similarity threshold, and constructing a clustering connection graph, we can ensure that nodes with connecting edges are likely to belong to the same category, while nodes without connecting edges are likely to not belong to the same category.
This improves the quality of the clustering connection graph, ensuring that nodes with connecting edges are likely to belong to the same category, while nodes without connecting edges are likely to not belong to the same category, thus improving the accuracy of the clustering connection graph.
Smart Images
Figure CN115661494B_ABST