With the invention of low-cost smartphones and other image capturing devices, image acquisition is no longer a difficult task. This has created a huge amount of unorganized images which require proper indexing for easy access. The field of Content-Based Image Retrieval (CBIR) proposes algorithms to solve this problem. This paper proposes a new multiresolution descriptor, named Multiscale Local Spatial Binary Gaussian Co-occurrence Pattern (MLSBGCP) for CBIR. Grayscale image is subjected to three-level Gaussian filtering process to perform multiresolution processing of the image. Local Spatial Binary Pattern (LSBP) of resulting filtered image is computed to gather local features. Finally, a feature vector is constructed using Gray-Level Co-occurrence Matrix (GLCM) which is then utilized as the feature vector to retrieve visually similar images. Performance of the proposed method is tested on Corel-1 K dataset and measured in terms of precision and recall. The experimental results demonstrate that the proposed method achieves better retrieval accuracy than some of the other state-of-the-art CBIR methods.
CITATION STYLE
Srivastava, P., & Khare, A. (2018). Content-Based Image Retrieval Using Multiscale Local Spatial Binary Gaussian Co-occurrence Pattern. In Lecture Notes in Networks and Systems (Vol. 19, pp. 85–95). Springer. https://doi.org/10.1007/978-981-10-5523-2_9
Mendeley helps you to discover research relevant for your work.