A hybrid structural feature extraction-based intelligent predictive approach for image registration

Based on characteristics or similarity metrics, image registration aligns various images acquired at different time frames or from different points of view. This is a common issue in image processing applications where it is necessary to do a parallel or collaborative analysis of two or more images of the same scene captured by different sensors, or images captured by the same sensor at various times. In this paper, we present an intelligent framework that uses a proposed hybrid structural feature extraction technique to estimate the input image’s transformation parameters using a ground truth image. This hybrid feature extraction technique extracts a few potential structural features as well as the object’s shape. The proposed technique’s effectiveness is compared with various prior state-of-the-art methods on publicly available image databases like SAR (synthetic aperture radar), medical, and standard benchmark images. The results of the parameter estimation experiments reveal that this hybrid feature extraction technique can reduce incorrect image registration parameter predictions in terms of mean square error, peak signal-to-noise ratio, root mean square error, average absolute intensity difference, and normalised cross-correlation coefficient and construct a good intelligent predictive framework.