纸浆纤维的形状特征提取应用研究

为了正确描述纸浆纤维图像的弯曲特性，提出了一种基于角点检测的纤维特征提取算法.本算法应用自适应滑动窗口策略进行纤维图像各边界点局部支持域的选取，以提取最优邻域，根据各边界点对应特征向量，将二维平面曲线转化为一维特征曲线，利用一维小波变换提取特征曲线奇异点，即二维平面曲线上的角点，根据边界点对应特征值与一维特征曲线的奇异特征确定纤维扭结个数，并根据各点对应局部支持域确定曲度与扭结平均角度，从而完成纸浆纤维的弯曲特征提取.实验结果证明，本算法能有效提取纸浆纤维弯曲特征.

Applied study on paper fibre shape characteristics extraction

A shape characteristics extraction algorithm based on corner detection was proposed to describe the bent characteristics of paper fibre image.Adaptive sliding window strategy was applied to select the optimal support field of boundary point;the two-dimension planar curve was transformed into one-dimension curve based on the eigenvectors of boundary point covariance matrices,and wavelet transform was utilized to extract the singularities of the one-dimension curve comprising the corners on the two-dimension planar curve,the number of kinks could be obtained according to eigenvalue of boundary point and singularities of one-dimension curve,the curl index and kink index could be obtained according to the support field of boundary point,so that the bent characteristics of paper fibre were described easily.The algorithm can extract fibre shape characteristics.Experiment results showed the method's validity and effectiveness.