Semantic feature selection for object discovery in high-resolution remote sensing imagery

Given its importance, the problem of object discovery in High-Resolution Remote-Sensing (HRRS) imagery has been given a lot of attention by image retrieval researchers. Despite the vast amount of expert endeavor spent on this problem, more effort has been expected to discover and utilize hidden semantics of images for image retrieval. To this end, in this paper, we exploit a hyperclique pattern discovery method to find complex objects that consist of several co-existing individual objects that usually form a unique semantic concept. We consider the identified groups of co-existing objects as new feature sets and feed them into the learning model for better performance of image retrieval. Experiments with real-world datasets show that, with new semantic features as starting points, we can improve the performance of object discovery in terms of various external criteria.