CAJ | 학술논문

为提高土地督察线索获取的快捷性和准确性，提出了一种基于稳定地物样本库的建设用地变化监测方法。首先，利用先验知识结合本底影像选取土地利用类型未发生变化的对象，形成稳定的地物样本库；然后，关联最新影像，动态筛选训练样本，实现建设用地变化自动监督分类；最后，基于先验知识对分类结果自动甄别，进行双重辅助变化信息提取，并利用相隔4个月的资源3号卫星数据开展对比试验。结果表明，该方法提取的建设用地变化位置准确率达72.82%(常规矢量监测法仅为33%)，大大提高了违法用地位置信息发现的准确率，具有较高的可靠性和精度。
위제고토지독찰선색획취적쾌첩성화준학성，제출료일충기우은정지물양본고적건설용지변화감측방법。수선，이용선험지식결합본저영상선취토지이용류형미발생변화적대상，형성은정적지물양본고；연후，관련최신영상，동태사선훈련양본，실현건설용지변화자동감독분류；최후，기우선험지식대분류결과자동견별，진행쌍중보조변화신식제취，병이용상격4개월적자원3호위성수거개전대비시험。결과표명，해방법제취적건설용지변화위치준학솔체72.82%(상규시량감측법부위33%)，대대제고료위법용지위치신식발현적준학솔，구유교고적가고성화정도。
In order to improve the rapidity and correctness of the clue acquisition method for land supervision, this paper proposes a method of automatic construction land change detection based on stable sample database. Stable sample database was constructed by using historical land inventory data. Then the images were associated to dynamically sieve training data so as to tremendously reduce human intervention for the sake of automatic supervised classification. Before change detection, the classification results were discriminated by land inventory data. ZY-3 data obtained 4 months apart were used to carry out comparative experiments. The accuracy of construction land change detection using the method proposed in this paper was 72 . 82%, while that of the traditional method only reached 33% on equal terms. The results show that the precision of illegal construction land change detection is highly improved.