中文摘要：

针对高分一号卫星（GF-1）玉米田遥感图像中玉米田光谱复杂和地块边缘模糊导致的面积统计误差大的问题,本文提出一种块模糊增强和最小值边缘提取相结合的边缘检测方法进行玉米田地块分割处理,以减小面积统计误差。首先将彩色遥感图像从RGB变换到I^1I^2I^3彩色空间,提取出含丰富特征的单色图I1;然后利用模糊理论对I1进行基于块的增强处理;再对增强后的图像进行最小值边缘提取;最后利用Full Lambda-Schedule算法对区域边缘进行优化。通过与Canny和Sobel等边缘提取方法比较,证明本文的边缘检测结果能有效地分割出玉米田地块目标,减少了玉米田光谱复杂和边缘模糊带来的影响,检测出的边缘更符合玉米田实际分布,玉米田面积统计结果更符合实际。

英文摘要：

This paper proposed a corn field segmentation method for GF-1 satellite remote sensing image based on blocky fuzzy enhancement and min edge extraction. This proposed method improves the accuracy of statistics of corn field area by solving the complexity of spectrum and reducing fuzziness of field edges. First, the color remote sense image was transformed from RGB format into I^1I^2I^3 format, and the monochrome figure I^1 which has rich characteristics was gotten. Then the image P was enhanced by blocks with fuzzy theory, and the edge of I^1 was extracted by using rain algorithm. Last, edge was optimized through Full Lambda-Schedule algorithm. Proposed method was compared with Canny and Sobel algorithm through experiments. The results showed that proposed method was effective in segmenting corn fields and detecting edge of color remote sensing image. The complexity and fuzziness were reduced effectively, and the results were more inline with the actual characteristics of corn distribution. Also the accuracy of statistics cornfield area was improved.