利用周期性结构纹理图像独特的频谱分布,设计出一种基于频谱特征的周期性结构参数检测方法,搭建了静态图像检测和实时采集检测系统,对织物密度进行多次检测实验。检测结果表明,相对误差小于1%的比例达到96.7%,平均误差比现有算法降低了约36%,单帧图像检测时间约为150 ms,适用于工业检测场景。将所提方法应用于纸张计数、木材纹理检测、微结构阵列检测等实际应用领域,实验结果表明,基于频谱特征的周期性结构参数检测方法的准确度较高,适用范围广,检测效率高。

A periodic structural parameter inspection method based on spectrum characteristics is designed herein by utilizing the unique spectrum distribution of periodic texture images. A static image inspection interface and a real-time acquisition and inspection system are built, and multiple experiments are conducted based on fabric density inspection. The ratio of relative error of experimental data less than 1% is 96.7%, the average error is about 36% lower than that of the existing algorithm, the inspection time of single frame image is approximately 150 ms, and the inspection speed is suitable for industrial inspection. The proposed method is applied to the field of paper counting, wood texture inspection, and micro-structure array inspection. The experimental results show that the periodic structure parameter inspection method based on spectrum characteristics has high accuracy, wide application range, and high detection efficiency.