基于局部加权拟合算法的无人机遥感影像多尺度检测技术

Multi-scale detection technology of UAV remote sensing image based on local weighted fitting algorithm

摘要:

目前提出的无人机遥感影像多尺度检测技术平均图像灰度较差，导致检测结果清晰度较低。为了解决上述问题，基于局部加权拟合算法研究了一种新的无人机遥感影像多尺度检测技术，选用最小二乘法进行多次循环计算，确定周围区域重复率，通过抽稀处理提高数据精度。根据高斯金字塔得到n阶的影像序列，利用高斯金字塔和差分尺度划分完成遥感影像的特征提取。引入加权拟合算法，构建有效影像数据集，确定影像网络模型，从而完成合并，实现影像数据的检测。实验结果表明，基于局部加权拟合算法的无人机遥感影像多尺度检测技术能够有效提高平均图像清晰度，增强检测结果的清晰度。

Abstract:

The currently proposed multi-scale detection technology for UAV remote sensing images has a poor average image gray level, resulting in low resolution of detection results. In order to solve the above problems, a new multi-scale detection technology of UAV remote sensing images is studied based on the local weighted fitting algorithm. The least square method is used for multiple loop calculations to determine the repetition rate of the surrounding area and improve the data accuracy through thinning processing. According to the Gaussian pyramid, the n-order image sequence is obtained, and the feature extraction of the remote sensing image is completed by using the Gaussian pyramid and the difference scale division. Introduce a weighted fitting algorithm, construct an effective image data set, determine the image network model, thus complete the merger, and realize the detection of image data. The experimental results show that the UAV remote sensing image multi-scale detection technology based on the local weighted fitting algorithm can effectively improve the average image sharpness and enhance the sharpness of the detection results.