Algorithms for Histogram Equalization in Image Enhancement and Link Prediction in Social Networks

Abstract

Data processing method is exploited to obtain the expected results by processing and analyzing the input data. In this research, we have focused on processing image and social network data in the area of image processing and social network analysis respectively by using data processing method. Gray-level context-driven histogram equalization and community-adaptive link prediction are proposed for image processing and social network analysis respectively. The abstractions of these two data processing methods are as follows. First, histogram equalization, which redefines the distribution of gray-levels in an image, is an important step in image processing to enhance the image quality. Until now, numerous histogram equalization techniques have been proposed, among which the majority of them have focused on solving the problem of how the gray-levels in the histogram of an input image should be properly partitioned so that the image produced by collecting all equalization results for the partitioned sub-histograms leads to the quality enhancement of the image. However, the partition based equalization methods have an inherent limitation of not being able to equalize a sub-histogram crossing a partition boundary, which is the main cause of image distortion. In this work, we propose a gray-level context-driven histogram equalization method to overcome this limitation. In short, rather than constraining disjoint mapping ranges of the gray-levels among the partitions, we devise two enabling techniques: (1) a mapping range for each gray-level with no range-disjoint constraint and (2) a mapping distance between two adjacent gray-levels to make a full exploitation of mapping flexibility of gray-levels. We formulate the histogram equalization problem integrating the two techniques into a flow optimization problem in a specially designed structure of a network, and solve it globally and efficiently. In addition, we seamlessly combine the factor of power consumption into our network flow optimization formulation to make an easy trade-off between image quality and power saving. Second, Link prediction is one of hot research topics in social network analysis. Link prediction problem is to find a small set of node pairs in the networks that are not directly connected, but will be very likely to be connected in the future. To improve the prediction accuracy, many works have attempted to consider the community information, if available, in the social network structure. One common strategy of the prior community-aware link prediction algorithms is that they devised a sort of unified link prediction formulation that simply includes a premium term to express whether a link is structurally in the same community or not. However, since the formulation of the premium term relies on the structural formation of communities only, it cannot take into account the fact that the communities in different social networks, though they form almost identical community structures, can make different levels of influence on the link prediction. To cope with this limitation, we propose an adaptive approach, in which we use two separate link predictions depending on inter or intra-links in community, and then balance the links based on the degree of community influence on link prediction. In conclusion, through experiments with the diverse datasets, it is shown that our proposed gray-level context-driven histogram equalization and community-adaptive link prediction are able to achieve much more improved performance compared to previous data processing methods in each of the image processing and social network analysis area.