Object-based image segmentation and retrieval for texture images

Abstract

In this paper, object retrieval techniques based on texture segmentation are proposed for use in local region texture images. In order to perform the segmentation of a local region texture image in this study, three texture features were first extracted: coarseness, contrast and directionality. For image retrieval, texture features for visual perception (TFVP), colour histogram for K-mean (CHKM) and shape features for principal moments of inertia (SFPMI) features of the object were extracted after segmentation. TFVP, CHKM and SFPMI are the texture, colour features and shape features, respectively, of the object region in an image; they are used for image retrieval. In order to more precisely represent the accuracy of this proposed method, the image segmentation/ image retrieval results are compared with other methods. It was shown that the segmentation performance of the proposed method evaluated using misclassification error (ME), relative foreground area error (RAE), modified Hausdorff distance (MHD) and AVE is better than other methods. The results clearly show that the performances of the proposed method for local region texture image retrieval are significantly superior to those of the other methods for global region texture image retrieval. The object segmentation in this study can accurately segment the image object, thereby obtaining image retrieval results with much higher precision.

Automated laser engraving system for the calibration and manufacturing of nonlinear scales for electrical measuring instruments

“…Functions of this package were used for dial images processing. 21,22 Image recognition and object detection algorithms are quite different. Some of them are very sensitive to various lighting conditions.…”