自适应差异多尺度形态学的风力机叶片红外图像增强研究

为提高采用红外检测技术检测风力发电机叶片内部缺陷的能力，改善红外热像图的清晰度低、背景光照不均和细节识别能力弱等问题，提出了一种自适应差异多尺度形态学(ADMM)的图像增强算法。在膨胀和腐蚀运算中分别选择不同尺度的结构元素作为操作对象，采用对比度改善系数比最大值作为多尺度及差异尺度选择的目标函数；利用开、闭及顶帽运算的优势，提取多尺度下图像的明暗细节信息；通过白顶帽与黑顶帽差值运算得到细节增强信息，并与原图像融合得到增强的红外图像；通过调整图像的灰度分布达到最佳的视觉效果。仿真实验结果表明，该算法不仅有极强的鲁棒性，而且能有效地增强光照不均情况下的红外图像细节，提高红外检测在风力发电机叶片中的探伤能力。

Study on Infrared Image Enhancement of Wind Turbine Blades Based on Adaptive Differential Multiscale Morphology(ADMM)

In order to improve the ability of infrared detection of internal defects of wind turbine blades and overcome the problems of low definition of infrared thermal images, uneven background illumination, and weak ability of detail recognition, an image enhancement algorithm was proposed based on ADMM. Different scale structural elements were selected as the operation objects in the expansion and corrosion operations, and the maximum contrast improvement coefficient ratio was used as the objective function of multi-scale and difference scale selection. The light and dark details of multi-scale images were extracted by using the advantages of open, close, and top-hat operations. Then, the differences between the white cap and the black cap were calculated to get the detail enhancement information, and the enhanced infrared images were obtained by fusing with the original images. Finally, the grayscale distribution of the image was adjusted to achieve the best visual effectiveness. Simulation results show that the algorithm has strong robustness and may effectively enhance the infrared image details under uneven illumination, improve the detection ability of infrared detection in wind turbine blades.