为解决探地雷达（GPR）三维图像（C-scan）信息维度低、成像效果主观等问题，以图像像素值的有效测量信息阈值及信息映射函数为研究对象，建立GPR C-scan的成像优化策略。通过以图像的SSIM和PSNR为评价指标，以室内钢筋混凝土墙体和道路路面为试验案例，量化分析变成像参数的可视化效果。试验结果表明，适合城市典型应用场景的GPR信息可视化方法包括映射阈值为70%及投影函数为平方函数。此外，提出融合振幅和频率的属性成像方法，丰富C-scan图像信息，组合信息可视化方法，实现准确表现目标物体几何形状并缓解噪声干扰的GPR
C-scan成像，有助于减少人为判读误差，提高C-scan解译的可靠性。
Ground
penetrating radar(GPR) threedimensional imaging, known as Cscan imaging, often suffers from limited information dimensionality
and subjective image quality. In this study, the authors explore two critical
imaging parameters, the valid measurement threshold and information
transformation function, to formulate an optimization strategy for GPR Cscan imaging. Furthermore, the authors assess image quality using
structural similarity and peak signaltonoise ratio as evaluation metrics. To quantify the effectiveness of
different imaging parameters, a case study is conducted involving indoor
reinforced concrete walls and road surfaces. The experimental findings indicate
that, for typical urban GPR applications, an optimal measurement threshold of
70% coupled with the mean method as the transformation function provides the
best results. Moreover, an attribute imaging method that combines amplitude and
frequency is proposed to enhance the information within Cscan images. This integrated visualization approach accurately
portrays the geometric characteristics of target objects while reducing noise
interference in GPR Cscan imaging. These enhancements
are advantageous in minimizing human interpretation errors and enhancing the
overall reliability of Cscan interpretation.
