示温漆图像阴影去除算法的研究与实现

Study and implementation of shadow removal method for thermal paint images

为了消除阴影对温度识别的精度及对等温线划分的准确度造成影响,提出了一种快速去除阴影的算法。把3种颜色特征不变性在示温漆图像上进行阴影检测的效果进行对比,选取应用效果最好的颜色特征不变性进行阴影检测,用数学形态学及阈值的处理方法准确地定位出阴影边界;然后在RGB颜色空间内,利用邻近的非阴影区域按照映射策略对各个独立的阴影区域进行阴影补偿,达到阴影去除的目的。实验结果表明,该方法不仅能快速的检测出阴影,而且在去除阴影的同时最大程度地保持了图像的原始信息。

To eliminate the effect on accuracy of temperature recognition and isotherm division with shadow,a fast shadow removal algorithm is proposed.First comparison of three kinds of invariant color models is carried on thermal paint images and the best invariant color model is selected to detect shadows.Morphological algorithm and threshold are used to get accurate shadow boundary.Then each shadow region is respectively compensated using its adjacent no-shadowed region obeying a certain mapping strategy in RGB color space to achieve the purpose of removing the shadow.Experiment shows that the method can not only quickly detect the shadow,but also remove the shadow of the image while maintaining the original information. 【