









摘 要:隨著我國城鎮化進程的不斷深入,大量鄉村居民地已經轉化為城鎮居民地,為了度量二者實時的轉化情況,該文提出一種基于監督分類機制的居民地分類方法,該方法首先采用邊緣特征及高斯函數量化影像上的局部特征,然后構建5種城鎮及鄉村居民地分類規則,其次創建訓練樣本對各類規則進行學習,最后通過城鎮及鄉村測試樣本驗證該文方法的精度。實驗表明,該文方法可以對高分辨率遙感影像城鎮及鄉村居民地進行初級分類,為“城鎮化”進程提供一個新的衡量指標。
關鍵詞:高分辨率遙感影像;居民地;監督分類;分類規則;訓練樣本
中圖分類號:P237 文獻標志碼:A 文章編號:2095-2945(2024)29-0154-04
Abstract: With the deepening of urbanization in China, a large number of rural residential areas have been transformed into urban residential lands. In order to measure the real-time transformation of the two, this paper proposes a residential land classification method based on supervised classification mechanism. Firstly, the edge feature and Gaussian function are used to quantify the local features on the image, and then five classification rules of urban and rural residents are constructed. Secondly, this paper creates training samples to learn all kinds of rules, and finally verifies the accuracy of this method through urban and rural test samples. The experimental results show that this method can classify urban and rural residential areas in high-resolution remote sensing images, which provides a new measurement index for the process of "urbanization".
Keywords: high-resolution remote sensing images; residential area; supervised classification; classification rules; training samples
我國已從“十五”時期開始推行“城鎮化”[1],人口、產業不斷向城鎮聚集,城鎮數量和規模得到大幅提升?!俺擎偦边^程致使大量鄉村居民地轉化為城鎮居民地,準確詳實地厘清掌握城鎮及鄉村居民地的空間變化信息,才能進一步度量兩者的轉化率,進而為政府部門在評估城鎮、鄉村土地資源及人口遷移發展變化提供重要數據支撐。現有文獻居民地變化提取研究方法,大多將鄉村居民地和城鎮居民地定義為統一整體進行提取,未進行分類提取,研究同一區域不同時期居民地分布情況為主。Li等[2]采用SIFT特征算子檢測居民地變化情況。Tang等[3]通過分析兩期遙感影像的局部特征差異度和相似度提取居民地變化信息,該方法檢測結果比較優良,但算法較為耗時。……