Résumé
Objective To extract Xinjiang Uyghur medicine image features and analyze the features; To investigate the image classification effect of the researched features; To find the suitable features for Xinjiang Uyghur medicine image classification; To lay the foundation for content-based medical image retrieval system of Xinjiang Uyghur medicine images.Methods The flowers and leaves of Xinjiang Uyghur medicine were treated as the research objects. First, images were under preprocessing. Then color and textural features were extracted as original features and statistics method was used to analyze the features. Maximum classification distance was used to analyze the main features obtained from image classification. At last, the classification ability of features was evaluated by Bayes discriminant analysis.Results Color and textural features were selected and classified. The correct classification rate of flower images was 85% and the correct classification rate of leaf images was 62%. The classification effect of flower images used by selected features was better than classification effect of original feature.Conclusion Compared with the classification of original features, the classification accuracy of flower medicine is higher through selected features. This research can lay a certain foundation for the further researches on Xinjiang Uyghur medicine images and the improvement of feature extraction methods.
Résumé
Iris image quality evaluation plays a very important part in iris computer recognition. An iris image quality evaluation method was introduced into this study to distinguish good image from bad image caused by pupil distortion, blurred boundary, two circles appearing not concentric, and severe occlusion by eyelids and eyelashes. The tests based on this method gave good results.
Sujets)
Humains , Amélioration d'image , Méthodes , Interprétation d'images assistée par ordinateur , Iris , Reconnaissance automatique des formes , Méthodes , Photogrammétrie , Méthodes , Normes de référenceRésumé
In present, the most basically used parameters for speaker identification are linear predictive coding (LPC) parameter, Mel frequency cepstrum coefficient(MFCC), etc. First in this paper only MFCC was used as the parameter and then Lempel-Ziv Complexity was combined with MFCC as parameters. The text-dependent recognition rate of 50 speakers increased from 42% to 80% and the text-independent recognition rate of 50 speakers increased from 60% to 72%. This test shows that Lempel-Ziv complexity, as a new parameter, can be applied to speaker identification.
Sujets)
Femelle , Humains , Mâle , Algorithmes , Intelligence artificielle , Dynamique non linéaire , Reconnaissance automatique des formes , Méthodes , Traitement du signal assisté par ordinateur , VoixRésumé
23 subjects' 8-lead (Fp1, Fp2, Cp3, Cp4, T7, T8, P7, P8) electroencephalogram (EEG) was recorded when they were doing mental arithmetic at four different levels. We calculated the information transmission time series in human cerebral cortex basing on EEG, and the Lempel-Ziv complexity and C1C2 complexity of these time series. When 20 subjects were doing the most difficult mental arithmetic, the information transmission series between lead at left-brain (Cp3, T7, P7) and other leads was of more complexity than others; a light "cross" could be seen after the information transmission matrix was converted to image; when complexity was calculated, the difference was more significant by use of C1 complexity than by other complexity measures.