Improved Image Fusion Algorithms Using Spatial Frequency for Remote Sensing Applications

Abstract

Remote Sensing (RS) is a branch of science in which information about an object, phenomena, or an area is acquired without having any physical contact using sensors that are usually kept either at satellites or aircraft. The interaction of electromagnetic radiations emitted from the sensor with objects, Earth s surface, atmosphere, etc. extract not only the distance information but also the details like polarization, wavelength, direction, etc. of the reflected wave that characterizes the object or phenomena under investigation. newlineThe quality of the images provided by the sensors is usually represented with respect to spatial and spectral resolutions. Based on the type of resolutions, sensors that are commonly used in RS applications can be classified into two, Panchromatic (PAN) sensor and Multispectral (MS) sensor. PAN sensor has the ability to produce an image with superior spatial quality but poor spectral quality, while MS sensor provide excellent spectral quality image with poor spatial quality. This complementary behaviour of sensors has been a big hurdle in RS applications for many decades. There are many RS applications which require high spatial quality and spectral quality images. Several bottlenecks that exist in sensor technology like bandwidth, limitation in storage, cost, etc. have limited the development of sensors capable of producing high-resolution images both with respect to spatial and spectral domain. newline However, engineers and scientists have achieved this feat by implementing algorithms capable of combining PAN and MS images into a single image with excellent spectral and spatial quality. This technique is known as image fusion. Image fusion can be defined as a technique of combining two or more complementary images in order to obtain a single image which contain all the goodies of all the images. The quality of resultant image is indeed dependent on fusion rule and image fusion algorithm. newline