China
Africa
Explore chapters and articles related to this topic
Eye-Centric ICT Control
Published in Philip D. Bust, Contemporary Ergonomics 2006, 2020
Fangmin Shi, Alastair Gale, Kevin Purdy
The above approach makes the complexity of 3D object recognition and location reduced to 2D object recognition only. Algorithms can then be developed and applied to the scene camera output to try to recognise objects in the scene. An efficient and reliable object identification method is under development in the research project. This performs image feature matching between a scene image and an image database that collects images of target objects. The image feature detection algorithm is based on the SIFT-Scale Invariant Feature Transformation, approach proposed by Lowe (2004). SIFT features are adopted because they have advantages over other existing feature detection methods in that their local features provide robustness to change of scale and rotation and partial change of distortion and illumination. An example showing the SIFT matching result for identifying an electric fan in a scene is given in Figure 3. At the right of Figure 3 is the image, which contains an electric fan, taken by the scene camera. The image of the electric fan to the left of the figure is one of the reference images in the database. The lines between the two images indicate the points of matching between the reference and the real object images and shows how an environmental object is recognised.
Introduction to Visual Computing
Published in Ragav Venkatesan, Baoxin Li, Convolutional Neural Networks in Visual Computing, 2017
SIFT is an algorithm for detecting feature points (customarily called SIFT features) that are supposedly invariant to changes in scale and rotation, or have a slight appearance change due to varying illumination or local geometric distortion. The algorithm first creates a multiscale representation of the original image. Then it finds the extrema in the difference of Gaussians in that representation and uses them as potential feature points. Some criteria are used to discard any potential feature point that is deemed a poor candidate (e.g., low-contrast points). Finally, a descriptor will be computed for any of the remaining points and its neighboring region. This descriptor (and its location in an image) is basically what we call a SIFT feature. The full algorithm as described in Lowe (1999) also covers other implementation issues related to fast matching of objects based on SIFT features.
Image Retrieval Using Keywords: The Machine Learning Perspective
Published in Spyrou Evaggelos, Iakovidis Dimitris, Mylonas Phivos, Semantic Multimedia Analysis and Processing, 2017
Zenonas Theodosiou, Nicolas Tsapatsoulis
SIFT features were originally proposed for object detection and recognition tasks. In these tasks a dedicated matching scheme is used to compare images or image regions. In keyword modeling this is not the case. The SIFT feature vector feeds the keyword visual models to produce an output indicating whether or not the corresponding keyword can be assigned to the image corresponding to this input vector. This difference, along with the dimensionality reduction, which is applied to produce SIFT based vectors of fixed dimensionality, leads to deteriorating performance in image retrieval compared to other types of features, like the MPEG-7 descriptors [910].
Research status and prospect of visual image feature point detection in body measurements
Published in The Journal of The Textile Institute, 2022
Wenqian Feng, Yanli Hu, Xinrong Li, Yuzhuo Li
The SIFT algorithm was proposed by David Lowe in 1999, and then refined in 2004 (Lowe, 2004). SIFT is a widely used feature point recognition methods, and has been successfully applied to computer vision algorithms such as target detection, target tracking, and large-scale image retrieval (Acharya et al., 2018). The SIFT algorithm is divided into four steps: scale space extreme value detection, feature point location, determination of feature point direction, and construction of feature point descriptors. For the detection of the feature points of a 2D image of body measurements, it is only necessary to focus on the first two steps of the algorithm and locate the feature points. Using the Gaussian blur operation in different scale spaces, that is, a Gaussian kernel and the image convolution operation, the image interval points are sampled and an image pyramid is constructed. This Gaussian pyramid is then used to subtract the upper and lower layers in each group to obtain the Gaussian difference image, and then the difference-of-Gaussians (DOG) function is applied to detect and locate key points (Xu et al., 2019). The advantage of this algorithm is that it is not affected by the light level. In the case of complex, noisy scenes, it detects prominent feature points such as edge points, corner points, and points with different brightness values in bright and dark areas.
A Training-Free Approach for Generic Object Detection
Published in IETE Journal of Research, 2022
Bhakti V. Baheti, Sanjay N. Talbar, Suhas S. Gajre
We would like to compare this LSS based feature extraction approach with some of the state-of-art sparse feature point extractors like Maximally Stable Extremal Regions(MSER) [25], Scale Invariant Feature Transform(SIFT) [10] and Speeded Up Robust Features(SURF) [11]. MSER is an algorithm for blob detection in images and is mostly used in object recognition and stereo matching. SIFT and SURF feature detectors compute distinctive invariant local features. SURF is basically an efficient alternative to SIFT in terms of speed and robustness. We implemented these feature extractors and results are shown in the figure below. Figure 6(a) shows a sample image of bottle.Figure 6(b)–(d) shows locations of extracted features with MSER, SIFT and SURF, respectively. These algorithms basically detect local features within image like blobs, corners or interest points. They do not as such contain information about self similarity of these features and hence the object shape. On the other hand, our proposed feature extraction scheme generates those feature locations containing strucrural information as shown in Figure 6(e).
Content-based image retrieval: A review of recent trends
Published in Cogent Engineering, 2021
Ibtihaal M. Hameed, Sadiq H. Abdulhussain, Basheera M. Mahmmod
Scale-invariant feature transform (SIFT) is one of the most widely used local descriptors introduced by David Lowe (Low, 2004), which contains a detector and a descriptor for key points. SIFT is robust against image rotation and image scaling, but it performs poorly in matching at high dimensions and need a fixed-size vector for encoding to perform image similarity checking. In image retrieval, SIFT has two drawbacks: it uses a large amount of memory and has a high computational cost (Montazer & Giveki, 2015). Soltanshahi et al. proposed (Montazer & Giveki, 2015) a method for CBIR based on the use of SIFT and local derivative pattern (LDP) to construct the feature descriptor. To overcome its limitation of high memory usage and computation cost, the authors proposed two methods to reduce SIFT dimensionality. The proposed system was tested using four datasets and proved its high retrieval performance for images that contain an object. However, it needs to be improved for nature images.