面向猴脸的吸收马尔科夫链图像显著性检测

传统的基于吸收马尔科夫链进行图像显著性检测方法只能检测出与图像背景差异较大的目标,或者位于图像中心的显著目标.但通常情况下,被关注的目标并不具有这样的条件.提出了一种面向对象的吸收马尔科夫链的显著性检测算法,并将其应用于金丝猴面部的显著性检测中.算法在传统的吸收马尔科夫链进行图像显著性检测的过程中,引入惩罚因子,依据一定的先验信息来动态调整吸收时间.根据超像素块与目标色彩信息之间的差异对颜色权重进行相应的奖励或惩罚,以指引算法能够正确提取多个显著目标.实验表明:相对于传统算法,算法能够更准确地检测出被关注的显著目标,尤其在图像中含有多个关注目标时,效果更加显著.

Image saliency detection via absorbing Markov chain for monkeys'faces

Traditional saliency detection via absorbing Markov Chain can only detect the objects which have great contrast to the background or in the center of the images.However,in fact,the objects human focus on do not usually have these features.To alleviate this problem,propose an object-oriented saliency detection algorithm via absorbing Markov and apply it to saliency detection for monkeys' faces.This algorithm introduces a penalty factor to traditional saliency detection via Markov Chain to dynamically adjust the absorbing times relying on some prior knowledge.The reward or punishment to the color weights is made according to the differences between the colors of the superpixel points and color information of objects to detect salient objects correctly.The experimental results demonstrate that comparing with traditional algorithm,the proposed algorithm can detect the objects which are focused on correctly,and especially when there are more than one object.