Image enhancement is an important pre-processing step for various image processing applications. In this paper, we proposed a physiologically-based adaptive three-Gaussian model for image enhancement. Comparing to the...Image enhancement is an important pre-processing step for various image processing applications. In this paper, we proposed a physiologically-based adaptive three-Gaussian model for image enhancement. Comparing to the standard three-Gaussian model inspired by the spatial structure of the receptive field (RF) of the retinal ganglion cells, the proposed model can dynamically adjust its parameters according to the local image luminance and contrast based on the physiological findings. Experimental results on several images show that the proposed adaptive three-Gaussian model achieves better performance than the classical method of histogram equalization and the standard three-Gaussian model.展开更多
文摘Image enhancement is an important pre-processing step for various image processing applications. In this paper, we proposed a physiologically-based adaptive three-Gaussian model for image enhancement. Comparing to the standard three-Gaussian model inspired by the spatial structure of the receptive field (RF) of the retinal ganglion cells, the proposed model can dynamically adjust its parameters according to the local image luminance and contrast based on the physiological findings. Experimental results on several images show that the proposed adaptive three-Gaussian model achieves better performance than the classical method of histogram equalization and the standard three-Gaussian model.
文摘针对三支高斯混合聚类算法(three-way Gaussian mixture model,T-GMM)的阈值通常为人为设定,增加算法的不确定性的问题,本文中将阴影集思想融入三支高斯混合模型,提出一种基于阴影集的三支高斯混合聚类算法(three-way Gaussian mixture model clustering based on shadow sets,ST-GMM);ST-GMM算法先构造一个关于阈值的目标函数,再通过优化算法选取最优阈值。基于10个不同类型的UCI数据集的实验结果表明:ST-GMM算法不仅继承了T-GMM算法的特点,同时有效地降低了人为设定阈值的误差,聚类细节的刻画也更加准确。针对评价指标的测试进一步验证了ST-GMM算法具有良好的聚类性能。
