【摘要】 在纺织品纹样设计或来样加工中,由于试样摆放、拍摄角度或织物自身的原因等,无法绝对地保证横向水平和纵向竖直的状态,即使事先经过人工手动调整,织物图像的偏斜也无法避免。而且由于现有的大部分织物组织分析等算法都对图像倾斜较为敏感,倾斜角度的存在会给识别结果造成较大误差,检测速度也较缓慢。因此,有必要寻求一种准确修正织物图像的办法。经过对比分析Laplacian、Sobel和Canny三种算子的边缘检测效果,采用Canny算子检测灰度图像的边缘,通过Hough变换检测直线,找到倾斜角后采用仿射变换进行图像修正和裁剪。结果表明,此方法能减少确定阈值的步骤,自动得出横纵向倾斜角度,自动化效果好,可行性较高。更多还原

【Abstract】 In textile pattern design or sample processing, due to the sample placement, the shooting Angle or the fabric itself, it is impossible to absolutely ensure the horizontal and vertical state, even with the manual adjustment in advance, the fabric image deflection is still unavoidable.Moreover, most of the existing algorithms, such as fabric construction analysis algorithms, are sensitive to the image tilt.The existence of the tilt angle will cause a large error in the recognition results, the detection speed is also slow, so it is necessary to find a way to correct the fabric image.After comparative analysis of the edge detection effects of Laplacian, Sobel and Canny operators, the Canny operator is used to detect the edge of gray image, and the Hough transform is used to detect the line.After the tilt angle is found, the image is corrected by affine transform and trimmed.The results show that the method can reduce the steps to determine the threshold value and automatically obtain the angle of horizontal and vertical inclination.更多还原