【摘要】 本文提出玻璃污迹的直方图小波包检测方法 .该方法能以不同的分辨率快速、准确地进行图像污迹程度的检测 .小波包的高低频特征及求取直方图凸包面积残差法 ,既能大大降低了图像处理的计算工作量 ,又能够消除噪声及异样污迹 (杂物 )的干扰 .文章借助污浊因子提出了一个判别洁净度的量化概念 ,并在实验中得到验证 .针对擦洗作业实际 ,文章提出背景图像人工和自动初始化的方法 ,最后给出并讨论了一个实验的结果更多还原

【Abstract】 A novel method for detecting the window glass dirty image based on the wavelet histogram is presented. It can inspect the image rapidly and accurately. By this approach, using the feature of wavelet packages and square concavity difference of the histogram, it can not only reduce the workload of image processing, but can delete the noise and the odd interruption of dirty image in some extent. We propose the conception of dirty factor for judging the glass image clean or unclean by digitization. Its validity is tested by experience. For the real working situation of cleaning robot, a approach of background image processing in self driven and manual work is given, at last, the experimental result is discussed.更多还原