基于图像边缘和对比度的显著目标检测

为了对自然场景中各类显著目标进行准确的检测,提出了一种结合图像边缘和多尺度对比度信息的检测方法。首先对图像进行快速双边滤波,在对图像平滑去除复杂纹理的同时保留物体结构边缘信息,然后用直线检测算子进行边缘检测得到边缘显著图。再利用对比度检测算子计算图像的区域对比度和全局对比度得到对比度显著图。融合边缘显著图和对比度显著图得到最后的显著目标图。实验表明该方法准确率高于大多数现有的显著目标检测方法。

Edge and Contrast based Salient Object Detection

In order to detect all kinds of salient object in natural scene, we present a method which combined image edges and contrast information. First, we use fast bilateral filter to smooth image and then detect image edges with line detection template. At the same time, calculate the local and global contrast value on smoothed image. Finally, we combine edge saliency image and contrast saliency image into final saliency map. We provide quantitative evaluation which shows the proposed method outperforms many other state-of-art methods.