基于频域特征点提取的素色机织物密度识别算法

通过计算机图像处理技术对织物图像进行分析，利用傅里叶变换技术将织物图像从空间域转换到频率域，并采用对比度拉伸、高斯低通滤波等图像预处理手段来增强图像质量，然后使用阈值处理技术对织物频谱图进行处理，得到包含织物密度信息的特征点坐标。通过频谱图上特征点坐标和空间域周期的对应关系，经计算最终得到织物经纬密度。实验结果表明，织物密度识别最高误差为2.8%，此方法可以实现三原组织素色织物的经纬密度的自动识别。

Identification algorithm of plain woven fabric density via feature point extraction in frequency domain

The fabric image was firstly converted from the spatial domain to the frequency domain using the Fourier transform technique. And enhance image quality by using image pre-processing tools of contrast stretching and Gaussian low-pass filtering. With the threshold processing techniques, the feature points containing the fabric density information were obtained from the above spectrum of the fabric. And at last, the woven fabric density was identified from the feature points coordinates. The experimental results show that, the presented method worked well for the identification of the warp and weft density of Mihara plain fabric with a relative error of 2.8%.