视觉感知下的无人机拍摄图像最邻近帧质量增强

Quality enhancement of the nearest frame of UAV⁃captured image under visual perception

针对无人机拍摄图像邻近帧清晰度不达标的问题,为提升图像帧节点的质量水平,提出视觉感知下的无人机拍摄图像最邻近帧质量增强方法。以视觉感知模型为基础,确定定量感知参数取值范围,再通过求解回归判定向量的方式,实现对无人机拍摄图像的质量评价。根据邻近帧结构相似性损失表达式,计算像素亮度残差,联合相关参数指标,定义定量增强函数条件,完成视觉感知下无人机拍摄图像最邻近帧质量增强。实验结果表明,所提算法作用下的无人机拍摄图像分辨率最大值为180 ppi,满足提升无人机拍摄图像帧节点质量水平、解决图像清晰度不达标问题的实际应用需求。

A method of quality enhancement of the nearest frame of the UAV⁃captured image under visual perception is proposed to improve the quality level of image frame nodes.The ranges of quantitative perception parameters are determined on the basis of the visual perception model.And then the quality evaluation of UAV⁃captured images is achieved by means of solving the regression decision vector.According to the structural similarity loss expression of adjacent frames,the pixel brightness residual is calculated,and the quantitative enhancement function condition is defined by combining the relevant parameter indexes to complete the quality enhancement of the nearest frame of the UAV⁃captured image under the visual perception.The experimental results show that the maximum resolution of the UAV⁃captured image by the proposed algorithm is 180 ppi,which meets the practical application requirements of improving the quality level of the frame nodes of UAV⁃captured images and its image sharpness.