Key points are not available for this paper at this time.
With the rapid advancement of the Internet of Things (IoT), massive volumes of multimedia data are continuously generated by distributed sensing devices and edge nodes. Efficient and accurate image retrieval from such data has become a key component in enabling advanced IoT applications. However, the constraints of edge computing—including limited bandwidth, low power budgets, and heterogeneous hardware—pose significant challenges to conventional image retrieval schemes. To address these issues, this paper proposes a lightweight and effective Content-Based Image Retrieval (CBIR) framework optimized for cloud-enabled IoT environments. Specifically, this paper introduces a new multi-feature construction scheme that integrates the Color Granular Descriptor (CGD) for fine-grained color characterization, the Double-Radius Local Binary Pattern (DR-LBP) for enhanced local texture extraction, and the Lower-Order Polar Harmonic Fourier Moments (LPHFMs) for capturing global shape features with strong rotational and scale invariance. The proposed scheme achieves high retrieval precision with low computational cost, making it well-suited for deployment in resource-constrained IoT environments. Extensive evaluations conducted on widely used benchmark datasets—including Corel-1K, Corel-5K, Corel-10K, Oxford105K, and GHIM-10K—demonstrate the superior performance and robustness of the proposed method, validating its practical applicability to IoT scenarios.
Tao et al. (Wed,) studied this question.