改进的路面图像背景校正算法

Modified Background Correction Algorithms for Pavement Images

摘要:

针对CCD(charge coupled device)摄像机在拍摄路面图像时存在的背景光照不均问题,对传统的背景校正算法进行了改进,提出了改进的基准背景法和两种改进的背景平均校正法：减背景法和校正系数法.结合实例对上述方法进行验证,结果表明,改进的基准背景法运算简便,适用于任何路面图像的背景校正;减背景法抑制背景噪声的能力优于校正系数法,但后者校正的图像更符合人眼的观测习惯.另外,对背景校正后的图像进行图像分割的结果也表明,当路面图像背景亮度均匀时,大津法可以获得理想的图像分割效果.

Abstract:

Pavement images through charge coupled device (CCD) cameras always go with nonuniform background illumination,therefore,the paper presents an improved typical methods for background correction,including modified Cheng’s method and two modified average background methods,background subtraction and correction coefficient method.The methods mentioned are applied to real pavement distress images and the results show that modified Cheng’s method has advantage in simplified calculation and can be applied to any pavement image.Moreover,background subtraction has stronger ability to eliminate background noise compared to the correction coefficient method,but images corrected by correction coefficient method are more suitable to the observation of human eye.Besides,the result of segmentation after background correction shows that Ostu method can provide good segmented images when the background is uniform in pavement images.