Feature Evaluation with High-Resolution Images

Automatic hemolysis identification on aligned dual-lighting images of cultured blood agar plates

“…Therefore, we considered many suitable keypoint extraction and description approaches and selected the two we found to give best performance, which turned out to be SIFT [16] and KAZE [17] [18]. Our pre-selection proved to be in line with studies that consider wide benchmark and high resolution images [14].…”

“…We considered many of the most popular feature extraction [10] and description tools [11], [12]. Comparative studies highlight the application-dependency of their performance [13], [14], [9]. We then conducted rapid testing to select the most performing techniques in our context.…”

Accurate and robust feature-based homography estimation using HALF-SIFT and feature localization error weighting

“…The experimental procedure is as follows: First, to evaluate the performance of the proposed method and comparative methods on homography estimation, we adopt several image pairs acquired from the Oxford VGG dataset [27] and used in experiments from prior literature [28]. The Oxford VGG dataset includes viewpoint changes and scale changes, and the high resolution dataset from the literature [28].…”

“…The Oxford VGG dataset includes viewpoint changes and scale changes, and the high resolution dataset from the literature [28]. The ground truth of the homographies of each image pair was provided in these datasets.…”

Introducing higher order correlations to marginals' subset of multivariate data by means of Archimedean copulas

“…While analysing real-life multivariate data such as financial data, e-commerce data, biomedical data, audio signals or high resolution images [1][2][3][4][5], we have many features that carry valuable information. However, if a number of features is large, the computational cost of proceeding such data is high.…”