Remote Sensing Image Retrieval Based on Attribute Profiles

In recent years, many approaches based on mathematical morphology have been applied in remote sensing image processing. In the paper, we proposed to use attribute profiles (APs) for textural extraction to the problem of content-based image retrieval (CBIR). In particular, four different attributes of APs with multiscale were applied to spatial information description. Then, the obtained textural images were processed by Gabor filters to generate the amplitude and energy coefficients. Finally, the extracted features along with histogram and color-moments were used for retrieval. The performance of textural descriptors were evaluated based on the UC Merced Land Use -- Land Cover data set. The experimental results show that the proposed method performs better than the popular Gabor texture.