Image Significance Region Detection Based on Global Color Clustering and Contrast
Chao Jia, Guangyu Wu, Fanshu Kong · 2019
In this paper, an image significance region detection algorithm combining global color clustering and color contrast is proposed. Firstly, by clustering the image color, the color of each pixel in the image is replaced by the color of the closest clustering center. Then, the clustering color contrast is used to calculate the significance region of the image, which enhances the significance region and reduces the influence of nonsignificant objects. Experimental results show that this method has higher precision and better recall rate, and significantly reduces the influence of complex texture on the calculation of significance region.