Abstract: This paper presents a preprocessing method for color-component compensation to improve the performance of traffic sign image processing.The RGB color-components are separated and candidate component is selected according to the chroma-features of traffic sign firstly.Then，threshold processing is applied on the candidate color-component to extract the sign region from background.A new traffic sign image is reconstructed by the replacement between threshold and its corresponding color-component.Finally，color extraction method is employed on the new image to figure out the candidate sign region accurately.Experimental results show that the proposed preprocessing method can improve the performance in traffic sign detection，especially in region location and shape checking.The proposed method can serve as a feasible solution for preprocessing of traffic sign image.

Key words: color-component compensation, image detection, preprocessing, component choice

摘要： 为提高图像预处理的效果，提出了一种基于颜色信息的颜色分量补偿预处理方法。该方法首先将实景图像分成RGB三个分量，根据要提取的交通标志颜色选择处理分量。接着对选择分量进行阈值处理实现候选标志与背景的粗分割。然后用阈值化的色彩分量替代原图中的对应分量重构为一幅新的彩色图像。最后通过颜色提取对重构图像进行处理以提取候选标志区域。实验表明，采用该方法进行交通标志的预处理，提高了交通标志检测时标志定位和形状检测步骤的准确度和速度，为交通标志预处理提供了一种可行的解决方案。

关键词: 颜色分量补偿, 图像检测, 预处理, 分量选择