Image Classification Using Graph Regularized Independent Constraint Low-Rank Representation

The low-rank representation (LRR), which has found extensive use in a variety of sectors, has proven to be superior at examining low-dimensional sub-space structures embedded in data. However, existing LRR algorithms do not take into account the influence of independence constraints, resulting in incomplete data structures. An innovative technique called image classification using graph regularized independent constraint low-rank representation (GRI-LRR) is developed in response to the aforementioned issues. This model can extract both the global and higher-order local structural information of the data, and these two structural information complement one another to increase the discriminative power of the matrix. Extensive testing on three benchmark face datasets and an object picture database demonstrates that the suggested strategy performs and is more reliable at classifying objects.