کد مقاله کد نشریه سال انتشار مقاله انگلیسی نسخه تمام متن
412700 679678 2010 9 صفحه PDF دانلود رایگان
عنوان انگلیسی مقاله ISI
Scene classification using multiple features in a two-stage probabilistic classification framework
موضوعات مرتبط
مهندسی و علوم پایه مهندسی کامپیوتر هوش مصنوعی
پیش نمایش صفحه اول مقاله
Scene classification using multiple features in a two-stage probabilistic classification framework
چکیده انگلیسی

Information in the frequency domain is useful in image classification. For natural scene classification, Oliva and Torralba proposed a global feature by sampling the power spectrum of the filtered image. In this paper, we present a hybrid global feature for scene classification. To capture the textural characteristics of the image in the frequency domain, we propose two feature extraction strategies based on gray-level co-occurrence matrices. Both contain statistics of the co-occurrence matrix, but the first one is of a much higher dimension than the second. We demonstrate that the proposed feature is a helpful supplement to the energy feature in terms of increased classification accuracy for real scene images. In order to combine these two kinds of features and further improve the classification accuracy, a posterior probability based two-stage classification method is proposed in which a linear combination of the probabilistic output of the SVMs in the first stage is used to train another SVM in the second stage. The overall results of the proposed method show an increase in the classification accuracy of about 1.5–4% compared to the original feature on four scene datasets. And the performance of the proposed method is also shown to be comparable to the recently reported results.

ناشر
Database: Elsevier - ScienceDirect (ساینس دایرکت)
Journal: Neurocomputing - Volume 73, Issues 16–18, October 2010, Pages 2971–2979
نویسندگان
, , ,