A DEEP CONVOLUTIONAL COUPLING NETWORK FOR CHANGE DETECTION BASED ON HETEROGENEOUS OPTICAL AND RADAR IMAGES

Abstract

We propose an unsupervised deep convolutional coupling network for change detection based on two heterogeneous images acquired by optical sensors and radars on different dates. Most existing change detection methods are based on homogeneous images. Due to the complementary properties of optical and radar sensors, there is an increasing interest in change detection based on heterogeneous images. The proposed network is symmetric with each side consisting of one convolutional layer and several coupling layers. The two input images connected with the two sides of the network, respectively, are transformed into a feature space where their feature representations become more consistent. In this feature space, the different map is calculated, which then leads to the ultimate detection map by applying a thresholding algorithm. The network parameters are learned by optimizing a coupling function. The learning process is unsupervised, which is different from most existing change detection methods based on heterogeneous images. Experimental results on both homogenous and heterogeneous images demonstrate the promising performance of the proposed network compared with several existing approaches.

Existing System

Kullback–Leibler divergence similarity, Dempster–Shafer fusion theory

Proposed System

This proposed system proposes a symmetric convolutional coupling network (SCCN) for change detection based on satellite images. There are three major characteristics of SCCN. First, the network has a symmetrical structure with each of the two sides composed of convolution and coupling layers for feature extraction and transformation. Second, feature transformation is made simultaneously from both sides of the network via hierarchical coupling layers by minimizing a coupling function, which measures the pixel wise difference summed over unchanged pixels. Third, the network parameters are first initialized by layer wise feature extractors incorporated with certain noise models of images to reduce the impact of noise and extract useful features, which facilitate the learning process. Different from the existing change detection methods based on heterogeneous images, the proposed method is fully unsupervised where unchanged pixels are automatically identified and utilized during the learning process.

Architecture

BLOCK DIAGRAM

ARCHITECTURE