自动化集装箱码头的图像增强算法研究
An image enhancement algorithm for automated container terminal
-
摘要: 为提高自动化码头应用场景拍摄图像的视觉质量,提升目标检测算法的准确率,提出一种结合改进的人工多曝光融合策略与白平衡处理的基于引导滤波分层处理(Guided Filtering Based Layering Processing, GFLP)的图像增强算法。利用改进的人工多曝光融合策略提升图像的对比度并重建图像细节;采用引导滤波公式对处理后的图像分别划分为基础层图像和细节层图像;对基础层图像进行“灰色世界”(Gray-Wond, GW)白平衡处理,将处理后的基础层和细节层进行简单相加得到增强后的图像。试验结果表明:与先进的图像增强算法相比,GFLP算法能更有效地提升自动化码头场景拍摄图像的视觉质量。该算法将有助于解决视觉技术实际应用的不足,对集装箱自动化码头的技术更新具有重要的意义。Abstract: The introduced image enhancement algorithm integrates the guided filtering-based layering processing(GFLP) with improved manual multi-exposure fusion and white balance processing. The improved manual multi-exposure fusion is used to enhance the image contrast and rebuild image details. The GFLP splits the image into basic image layer and detail image layer. The basic image layer is processed with “grey world” white balance. The processed basic image layer and the detail image layer are arithmetically added together to form an enhanced image. The advantage of the GFLP process over more sophisticate image enhancement algorithms for the particular application in automated container terminal is demonstrated.