W pracy opisano implementację oraz analizę eksperymentalną algorytmu Entropodynamicz-nego Filtra Percentylowego, pozwalającego na detekcję szumu w obrazach o wielu składowych spektralnych. Kostka danych wizualnych jest przetwarzana tak, aby wygenerować, niezależnie dla każdej składowej spektralnej, mapę krawędzi, która pozwala na oszacowanie informacji o rozkładzie entropii w spektrum. Filtr percentylowy oddziela nośniki szumu od warstw wysoce informacyjnych. Jakość metody jest weryfikowana dzięki serii testów wykonanych dla zadania klasyfikacji.
EN
Following work describes the implementation and experimental evaluation of the Entropodynamic Percentile Filter algorithm, allowing the detection of noise in images with many spectral components. The visual data block is processed to generate an edge map, independent of each spectral component, which makes possible the estimation of the information on the distribution of entropy in the spectrum. An appropriately constructed percentile filter separates noise carriers from highly informative layers. The quality of the method is verified with a series of experiments performed for the classification task.
An insufficient number or lack of training samples is a bottleneck in traditional machine learning and object recognition. Recently, unsupervised domain adaptation has been proposed and then widely applied for cross-domain object recognition, which can utilize the labeled samples from a source domain to improve the classification performance in a target domain where no labeled sample is available. The two domains have the same feature and label spaces but different distributions. Most existing approaches aim to learn new representations of samples in source and target domains by reducing the distribution discrepancy between domains while maximizing the covariance of all samples. However, they ignore subspace discrimination, which is essential for classification. Recently, some approaches have incorporated discriminative information of source samples, but the learned space tends to be overfitted on these samples, because they do not consider the structure information of target samples. Therefore, we propose a feature reduction approach to learn robust transfer features for reducing the distribution discrepancy between domains and preserving discriminative information of the source domain and the local structure of the target domain. Experimental results on several well-known cross-domain datasets show that the proposed method outperforms state-of-the-art techniques in most cases.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.