Image Fusion. Algorithms and Applications

Citation

(2009), "Image Fusion. Algorithms and Applications", Sensor Review, Vol. 29 No. 3. https://doi.org/10.1108/sr.2009.08729cae.001

Image Fusion. Algorithms and Applications

Article Type: Book review From: Sensor Review, Volume 29, Issue 3

Tania Stathaki,Elsevier,June 2008,ISBN: 978-0-12-372529-5,$130,520 p.,www.elsevier.com/wps/find/bookbibliographicinfo.cws_home/714207/description#bibliographicinfo

The major aim of this book is to provide a collection of recent advances in the field of image fusion. With the proliferation of image and sensor modalities an important problem is given by the integration of multiple sensorial information in order to create a multi-modal image that provides a higher level of information that can be used in the process of scene analysis and understanding. This book consists of a collection of 19 chapters that were authored by well-known researchers in the field of image fusion where both theoretical and practical issues related to the integration of multi-sensorial information were addressed in detail. The mathematical foundation of each chapter is solid and the proposed image fusion algorithms described in each chapter were validated using comprehensive comparative studies. This book is suitable for all computer vision researchers as it covers a large spectrum of theoretical and practical problems, but it is particular useful as a reference text for academic, industrial, and postgraduate researchers that work in the field of image fusion.

Chapters 1-3, 6, and 9 introduce the principles and the current trends in super-resolution (SR) image reconstruction where one of the main issues was the presentation of representative state of the art SR methods. In these chapters the authors discussed the common image degradation models and the techniques that are applied to reconstruct the SR image from multiple low-resolution images. Chapter 16 is also related to SR image reconstruction, where the authors have introduced the main concepts that are currently applied in the development of pan sharpening algorithms for satellite imagery.

Chapters 4, 5, 10, 12, and 14 address the image fusion of artificially distorted images (motion blur, out of focus) and the integration of the information generated by a standard monochrome camera and an infrared sensor. In Chapter 4 the authors developed an image fusion approach based on the analysis of the input images in the Independent Component Analysis domain. An alternative technique for adaptive integration of multi-modal images (monochrome and infrared) using an approach based on the analysis of images in the wavelet domain is detailed in Chapter 5. Chapter 11 deals with the problem of edge fusion performed on multi-modal images, where the main emphasis has been placed on the development of a statistical framework based on the calculation of Kappa coefficients. In Chapter 13 the authors describe the application of the empirical mode decomposition to tasks such as image restoration and multi-modal sensor fusion.

Chapter 7 provides a comprehensive study that analyses the application of Bayes theory to image fusion. The authors demonstrate that practical problems such as 3D reconstruction can be solved within the Bayesian framework, where the problem of data fusion can be re-formulated and the optimal result can be achieved by minimising the weighted sum of partial energies that model the fusion process.

Chapter 8 gives an interesting overview on several methods to extent the field of view (FOV) when the image data is captured in conditions that there is a relative motion between the camera and the static scene. They demonstrate that the addition of additional optics that are fitted to a standard CCD or CMOS sensor offers the possibility to construct panoramic images that are characterised by an extended FOV and depth of filed. In the same chapter, the authors also proposed the inclusion of a spatially varying filter to obtain panoramic images with intensity high dynamic range.

Chapter 15 evaluates the feasibility to apply image fusion algorithms to practical applications such as remote sensing and non-destructive testing. In their evaluation the authors investigated a number of image fusion techniques including multiple Kalman filtering, pixel level, feature level and symbol level (SL) and they conclude that the SL techniques produce more consistent results than the other techniques when applied for defect classification tasks.

The final Chapters 17-19 were concerned with the introduction of metrics to evaluate the quality of the fusion process and a comprehensive performance evaluation is provided for standard image fusion techniques.

Ovidiu GhitaVision Systems Group, School of Electronic Engineering, Dablin City University, Ireland