Adaptive multisensor target detection using feature-based fusion

Target detection techniques play an important role in automatic target recognition (ATR) systems because overall ATR performance depends closely on detection results. We propose a multisensor target detection technique based on adaptive feature-based fusion suitable to detect low-contrast and/or blurry targets with relatively low computational complexity. The proposed technique extracts four different features designed to find regions of high contrast, strong edges, or high information content. The proposed multisensor target-detection technique incorporates multiple imaging sensors in different spectral ranges, such as the visible, near-infrared, and far-infrared bands. The images from the multiple sensors are jointly processed via feature extraction and multisensor fusion. In the multisensor fusion process, a multisensor confidence image is created by adaptively combining the features generated from the multisensor images to obtain more accurate potential target locations. Experimental results on multisensor test sequences are provided.