China
Africa
Explore chapters and articles related to this topic
Semantic Enrichment for Automatic Image Retrieval
Published in Spyrou Evaggelos, Iakovidis Dimitris, Mylonas Phivos, Semantic Multimedia Analysis and Processing, 2017
Clement H. C. Leung, Yuanxi Li
Content-based image retrieval (CBIR) [265, 197, 221, 697, 979, 667, 636, 707, 414] is a technique used for retrieving similar images from an image database. CBIR operates on retrieving stored images from a collection by comparing features automatically extracted from the images themselves. The most common current CBIR systems, whether commercial or experimental, operate at level 1. A typical system allows users to formulate queries by submitting an example of the type of image being sought, though some offer alternatives such as selection from a palette or sketch input. The system then identifies those stored images whose feature or signature values match those of the query most closely, and displays thumbnails of these images on the screen. The most challenging aspect of CBIR is to bridge the gap between low-level feature layout and high-level semantic concepts.
Big Data in Computational Health Informatics
Published in Ayman El-Baz, Jasjit S. Suri, Big Data in Multimodal Medical Imaging, 2019
Ruogu Fang, Yao Xiao, Jianqiao Tian, Samira Pouyanfar, Yimin Yang, Shu-Ching Chen, S. S. Iyengar
Content-based image retrieval (CBIR) reveals its crucial role in medical image analysis by providing physicians and doctors with diagnostic aid including visualizing existing and relevant cases, together with diagnosis information. Therefore, retrieving images that can be valuable for diagnosis is a strong necessity for clinical decision-support methods including evidence-based medicine or case-based reasoning. In addition, their use will allow for the exploration of structured image databases in medical education and training. Therefore, information retrieval in medical images has been widely investigated in this community.
Image Retrieval
Published in Vipin Tyagi, Understanding Digital Image Processing, 2018
Content-based image retrieval (CBIR) systems are more intuitive and user-friendly in comparison to text-based image retrieval systems. A CBIR system uses the visual contents of the images which are defined using low-level features of image contents like color, texture, shape and spatial locations of the images in the databases. The CBIR system retrieves and outputs matched images when an example image or sketch is given as input into the system. Querying in this way eliminates the need for annotation in the images and is close to human perception of visual information.
Integrating content-based image retrieval and deep learning to improve wafer bin map defect patterns classification
Published in Journal of Industrial and Production Engineering, 2022
Ming-Chuan Chiu, Yen-Han Lee, Tao-Ming Chen
Because the standard WBM expert inspection procedure is time-consuming and labor-intensive [10,11], machine learning and deep learning methods have begun to be used during analysis to identify defect pattern types. However, while applying machine learning or deep learning methods to automatic defect pattern classification, problems have arisen because of insufficient training data for certain defect patterns, largely because of the increased yield rate and patterns distribution. To address the issue, this research has developed a method integrating content-based image retrieval (CBIR) with deep learning that can help obtain accurate identification and classification results with only a small amount data in a short time. CBIR, an image retrieval-and-classification technique, allows engineers to directly and efficiently utilize the data, and it reduces the quantity of data needed for classification model training. The use of deep learning techniques can further improve classification results, which then strengthens the recognition capability within more complex images. The proposed CBIR-CNN combination method can enable companies to improve their defect pattern classification analysis and to effectively increase yield with less time and effort.
Local mesh ternary patterns: a new descriptor for MRI and CT biomedical image indexing and retrieval
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2018
Health care services today rely heavily on diverse biomedical imaging data which have been expanded exponentially in quantity – as there is rapid increase in the number of medical check-ups per day; because of the use of diverse range of imaging modalities such as magnetic resonance imaging (MRI), ultrasound (US), computed tomography (CT), X-ray, etc. for different clinical studies. As we know that medical imaging is made up of dissimilar minor structures, there has been much interest of researchers in the development of well-structured techniques to work on huge image databases of biomedical images for efficient access, search and retrieval. To sort out the problem of medical images, the knowledge of the content-based image retrieval (CBIR) approach is disseminated to develop content-based medical image retrieval. Some comprehensive and extensive literature survey on CBIR systems is presented in Niblack et al. (1993), Antani et al. (2004), Jing and Allinson (2008), Sihyoung et al. (2010) and Yue et al. (2011). CBIR uses the visual content features such as colour, texture, shape and spatial layout of regions or objects to represent and index the biomedical image database for efficient retrieval. These features are arranged as multi-dimensional feature vectors and stored in the feature database. The main step of the CBIR is feature extraction, the effectiveness of which rests upon the method derived for extracting features from given images. The selection of feature descriptors affects on-image retrieval performance.
Multiple kernel scale invariant feature transform and cross indexing for image search and retrieval
Published in The Imaging Science Journal, 2018
B. Mathan Kumar, R. PushpaLakshmi
The image retrieval approaches can be classified into three: text-based, content-based [8,9], and semantic-based schemes [10,11]. In the text-based technique, the images are retrieved using keywords interpreted on images, which depends on images manually labelled with keywords [12]. The image search process that is performed based on its visual features, such as colour, texture, and shape, extracted automatically, is known as CBIR. Most of the approaches use Bag-of-Visual-Words [13] model utilizing invariant features to represent the image. Image representation is a visual word vector that is generally formed by clustering the obtained local features [14]. In Content-Based Image Retrieval (CBIR), the retrieving process is based on several low-level features such as textures, colour, shapes, or other information that represents an image. The applications of CBIR include medical application, crime provision, and fingerprint identification [15]. The semantic-based approach has become an important search area, which uses abstract features and image attributes. It bridges the semantic gap that exists between the descriptive power of low-level image feature and high-level semantic features [16].