A Coarse-to-Fine Registration Strategy for Multi-Sensor Images with Large Resolution Differences

Kai Li,Zhenchao Zhang,Yongsheng Zhang,Guangling Lai

doi:10.3390/rs11040470

Kai Li, Zhenchao Zhang + Show 2 more

Open Access

https://doi.org/10.3390/rs11040470

Copy DOI

Abstract

Automatic image registration for multi-sensors has always been an important task for remote sensing applications. However, registration for images with large resolution differences has not been fully considered. A coarse-to-fine registration strategy for images with large differences in resolution is presented. The strategy consists of three phases. First, the feature-base registration method is applied on the resampled sensed image and the reference image. Edge point features acquired from the edge strength map (ESM) of the images are used to pre-register two images quickly and robustly. Second, normalized mutual information-based registration is applied on the two images for more accurate transformation parameters. Third, the final transform parameters are acquired through direct registration between the original high- and low-resolution images. Ant colony optimization (ACO) for continuous domain is adopted to optimize the similarity metrics throughout the three phases. The proposed method has been tested on image pairs with different resolution ratios from different sensors, including satellite and aerial sensors. Control points (CPs) extracted from the images are used to calculate the registration accuracy of the proposed method and other state-of-the-art methods. The feature-based preregistration validation experiment shows that the proposed method effectively narrows the value range of registration parameters. The registration results indicate that the proposed method performs the best and achieves sub-pixel registration accuracy of images with resolution differences from 1 to 50 times.

Highlights

Image registration is the work of geometrically aligning two images containing the same scene, which are often called the reference image and the sensed image
Multiple sets of remote sensing image pairs acquired by multi-sensors at different times and different resolutions were used to evaluate the performance of the proposed method
It can be seen that large illumination differences and scene changes existed in the images

Summary

Introduction

Image registration is the work of geometrically aligning two images containing the same scene, which are often called the reference image and the sensed image. Image registration plays an important role in various applications, such as environmental monitoring, medical diagnosis, computer vision, and change detection [1,2,3,4]. For remote sensing image applications, the registration accuracy is of great concern. For applications like change detection, a registration accuracy of one-fifth of a pixel can result in a detection error of about 10%. Automatic registration algorithms provide a more practical means with high efficiency and accuracy and many methods have been proposed recently [6,7,8,9,10]. There are always problems like inefficiency, inaccuracy, and instability when it comes to the automatic registration of multi-sensor images. Further studies are required in order to improve the efficiency, accuracy, and robustness of the existing methods, especially for images with significant differences

Objectives

Methods

Results

Discussion

Conclusion