针对浅表层缺陷与正常区域特征混叠问题, 提出了一种基于遗传算法与时序红外热图加权叠加的红外无损检测方法。研究以铝板的 16类孔洞缺陷为对象, 采集预热试件降温过程的时序红外热图, 获取相应时序灰度图; 并以时序图中缺陷和正常区域灰度差值的加权和为目标函数, 采用遗传算法优化加权系数; 基于最优加权系数, 对时序灰度图依次进行加权叠加和梯度增强处理, 并对增强效果进行评估。结果表明: 经加权叠加和梯度增强处理后, 缺陷与正常区域的灰度比分别提升 8.5%和 31.0%。缺陷特征得到显著增强。

Aiming at the shallow surface defects and normal regional characteristics aliasing problems, the paper proposed a method of infrared nondestructive testing based on genetic algorithm and sequence infrared thermography weighted stack. The study focused on the object of 16 classes hole defects of aluminum plate. Firstly, sequence infrared image of aluminum plate during the cooling process were collected, the sequential grayscale were obtained from which. Secondly, gray difference weighted sum between defect and normal area was put in sequence diagram as objective function, and genetic algorithm was used to optimize the weighted coefficients. Then, the sequence gray diagram was weighted superposed and gradient enhancement was processed based on the optimization of the weighted coefficient, and then the effect of enhancement was evaluated. The results show that: after the weighted gradient overlay and enhancement processing, gray ratio between defects and normal area have been raised up to 8.5% and 31.0% respectively, and the defect feature are enhanced significantly.