Wyniki wyszukiwania - BazTech

Ograniczanie wyników

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: deskryptor cech

Sortuj według:

Ogranicz wyniki do:

Facial expression recognition under difficult conditions: A comprehensive study on edge directional texture patterns

Ahmed F., Kabir M. H.

International Journal of Applied Mathematics and Computer Science

2018

Vol. 28, no. 2

399--409

In recent years, research in automated facial expression recognition has attained significant attention for its potential applicability in human–computer interaction, surveillance systems, animation, and consumer electronics. However, recognition in uncontrolled environments under the presence of illumination and pose variations, low-resolution video, occlusion, and random noise is still a challenging research problem. In this paper, we investigate recognition of facial expression in difficult conditions by means of an effective facial feature descriptor, namely the directional ternary pattern (DTP). Given a face image, the DTP operator describes the facial feature by quantizing the eight-directional edge response values, capturing essential texture properties, such as presence of edges, corners, points, lines, etc. We also present an enhancement of the basic DTP encoding method, namely the compressed DTP (cDTP) that can describe the local texture more effectively with fewer features. The recognition performances of the proposed DTP and cDTP descriptors are evaluated using the Cohn–Kanade (CK) and the Japanese female facial expression (JAFFE) database. In our experiments, we simulate difficult conditions using original database images with lighting variations, low-resolution images obtained by down-sampling the original, and images corrupted with Gaussian noise. In all cases, the proposed method outperforms some of the well-known face feature descriptors.

Porównanie metod opartych na cechach 2D i 3D w zadaniu samolokalizacji robota mobilnego na podstawie danych RGB-D

Nowicki M.

Pomiary Automatyka Kontrola

2014

R. 60, nr 10

845--848

Praca porusza problem estymacji transformacji przestrzennej pomiędzy kolejnymi pozycjami robota mobilnego. Transformacja wyznaczana jest na podstawie danych RGB-D uzyskanych z sensora Kinect w kolejnych pozycjach wzdłuż trajektorii ruchu robota. Bardzo duża liczba punktów w przestrzeni otrzymanych z sensora Kinect powoduje, że nie mogą być one przetwarzane w czasie rzeczywistym na większości komputerów pokładowych robotów mobilnych. Dlatego w pracy porównano dwie metody ekstrakcji punktów charakterystycznych (kluczowych) redukujących rozmiar zbioru danych: cechy oparte na detekcji w chmurze punktów 3D oraz cechy oparte na detekcji na dwuwymiarowym obrazie RGB. Metody te porównano pod względem czasu działania, skuteczności ekstrakcji punktów oraz dryftu estymaty pozycji robota na publicznie dostępnych zbiorach danych RGB-D.

This paper addresses the problem of determining the egomotion between consecutive robot poses. The RGB-D Kinect sensor is used, which yields large amount of 3D points. It is impossible to process these data in realtime on most of the mobile robots. Thus, we present two approaches to point feature detection: 3D geometric features obtained from point clouds [9] and photometric 2D features detected in the RGB image [1]. Both methods are compared on publicly available RGB-D datasets [8, 10]. The detection on 2D image is the core of the currently state-of-the-art SLAM systems [3] while 3D features are an invention made especially for data captured by the Kinect-like sensors. First, it is demonstrated that both methods can be a part of a successful visual odometry system. Moreover, it is shown that detection of 2D image features is much faster than 3D, while the description is faster in case of 3D features (Tab. 1.). Performed experiments revealed that 3D geometric features tend to work better in environments of richer geometric structure (Tab. 2.) while method using photometric 2D features can be successfully expanded to additionally use 2D features detected on the depth image.