一种高空间分辨率的遥感变化检测方法在智慧农业中的应用

传统方法在确定影像对象的异质性时,根据整幅遥感影像以及判别经验所确定的全局固定阈值往往不能很好地适应各种不同属性的检测对象。针对这一问题,该研究提出了一种自适应的双模糊阈值的判别方法,在传统的图像变化检测预处理的基础上,利用Q型因子在整幅影像中获取具有代表性的训练样本,分别计算各样本的变化强度和相关系数的最优阈值以及熵的二值化阈值,建立样本的变化阈值集合,选择集合的中位数作为整幅影像的变化阈值,利用模糊识别算法分别对所得到的2幅变化影像进行运算,求交集建立混淆矩阵,最终得到变化检测的结果。试验结果表明,该算法对不同属性的影像对象具有良好的适应性,较传统的阈值变化检测方法其平均正确率提高了31.12%,有效地减少了错判或漏判。

The Application of Change Detection by High Spatial Resolution Remote Sensing in the Wisdom Agriculture

The traditional method that identifies global fixed threshold according to the experience and the whole image can not adapt to the different properties of detecting objects when determine the heterogeneity of the image object.Concerning this issue, this paper proposes a method based on adaptive double fuzzy threshold.Preprocessing finishes the prophase job to make it easier to do the following recognizing works, which includes binarization, smoothness and refinement such image standardization operations before the image change detection .Rep-resentative samples are obtained in accordance with Q factor in the whole image .There exists an optimal threshold index of change magnitude and the correlation coefficient, and binarization threshold of entropy to describe the change extent between two results.It is necessary to estab-lish a sample collection of change threshold, and make the median of a set as the change threshold of the whole image .Finally, change detec-tion results are obtained after counting intersection and establishing confusion matrix.Results show that the algorithm has a good adaptability for different image objects.Compared with traditional change detection method, the average accuracy of identifying is improved by 31.12%, which effectively reduces mistakes.