Abstract: Aiming to lines reconstruction and understanding in images, this paper puts forward maximum energy response by Laplace filter of Gaussian with multi-scale and multi-angle. It adaptively enhances the lines based on multi-scale space theory. Meanwhile it obtains lines scale, angle information and achieves lines detection using on-maximum suppression. Using lines scale and angle information acquired, the lines are reconstructed at last. The experiment results indicate the validity of lines reconstruction algorithm.

Key words: multi-scale analysis, lines detection, lines reconstruction

摘要： 针对图像中线条的重构与理解，基于多尺度空间理论，提出通过多尺度、多角度高斯-拉普拉斯滤波的最大能量响应，对线条进行自适应增强，同时获取线条的尺度、角度信息，利用非极大值抑制对线条实现检测，利用已获取的线条尺度与角度信息对线条进行重构。实验结果表明该线条重构算法的有效性。

关键词: 多尺度分析, 线条检测, 线条重构