【摘要】 针对传统的建筑物立面损毁自动检测方法存在检测时间过长、误检率较高等问题,提出一种基于基尼系数的建筑物立面损毁自动检测方法。将全变分去噪模型和分数阶微分理论两者相结合,构建新的建筑物立面图像去噪模型,对图像进行去噪处理得到平滑图像。采用Hear小波变换,在最高层对近似分量进行自适应阈值分割获取建筑物立面损毁初始区域。在其他各层中针对损毁区域进一步进行分割处理,通过基尼系数自动判定建筑物立面是否存在损毁。实验结果表明,所提方法能够有效抑制噪声干扰,减少检测时间,提高检测效率,获取准确的检测结果。更多还原

【Abstract】 Aiming at the problems of long detection time and high false alarm rate in traditional automatic detection method of building facade damage, an automatic detection method of building facade damage based on Gini coefficient is proposed. Combining the total variation denoising model with fractional differential theory, a new denoising model of building elevation image is constructed, and the smooth image is obtained by denoising the image. Hear wavelet transform is used to segment approximate components at the highest level by adaptive threshold to obtain the initial damage area of building elevation. In other layers, the damaged area is further segmented and the damage of the building facade is automatically determined by Gini coefficient. Simulation results show that the proposed method can effectively suppress noise interference, reduce detection time, improve detection efficiency and obtain accurate detection results.更多还原