کد مقاله کد نشریه سال انتشار مقاله انگلیسی ترجمه فارسی نسخه تمام متن
555870 874165 2016 10 صفحه PDF ندارد دانلود رایگان
عنوان انگلیسی مقاله
A targeted change-detection procedure by combining change vector analysis and post-classification approach
ترجمه فارسی عنوان
یک روش تغییر تشخیص هدفمند با ترکیب بررسی‌های برداری تغییر و رویکرد پس از طبقه بندی
کلمات کلیدی
تشخیص تغییر ؛ تشخیص تغییر هدفمند ؛ توصیف بردار پشتیبان دامنه ؛ همجوشی طبقه بندی
Change detection; Targeted change detection; Support Vector Domain Description; Classifier fusion
موضوعات مرتبط
مهندسی و علوم پایه مهندسی کامپیوتر سیستم های اطلاعاتی
چکیده انگلیسی

In remote sensing, conventional supervised change-detection methods usually require effective training data for multiple change types. This paper introduces a more flexible and efficient procedure that seeks to identify only the changes that users are interested in, here after referred to as “targeted change detection”. Based on a one-class classifier “Support Vector Domain Description (SVDD)”, a novel algorithm named “Three-layer SVDD Fusion (TLSF)” is developed specially for targeted change detection. The proposed algorithm combines one-class classification generated from change vector maps, as well as before- and after-change images in order to get a more reliable detecting result. In addition, this paper introduces a detailed workflow for implementing this algorithm. This workflow has been applied to two case studies with different practical monitoring objectives: urban expansion and forest fire assessment. The experiment results of these two case studies show that the overall accuracy of our proposed algorithm is superior (Kappa statistics are 86.3% and 87.8% for Case 1 and 2, respectively), compared to applying SVDD to change vector analysis and post-classification comparison.

ناشر
Database: Elsevier - ScienceDirect (ساینس دایرکت)
Journal: ISPRS Journal of Photogrammetry and Remote Sensing - Volume 114, April 2016, Pages 115–124
نویسندگان
, , ,