Data mining is “the science of extracting useful knowledge from huge data repositories” . Extending this definition, it can be concluded that image data mining is the science of extracting valuable knowledge from large volumes of images. The process of mining image data is theoretically simple. The first step is to extract features or properties from the source image, for example, water, trees, airplanes, clouds, cars, and so on. The second step is to use these features to classify the image within some class (also known as image classification) or to search in a database for similar images (better known as image retrieval). Image data mining is used in various fields: military reconnaissance, weather forecasting, agriculture, and even medicine, where medical images can be used in disease diagnoses.
This book presents the latest developments in the field of image data mining and is divided into four parts: “Preliminaries” (three chapters), “Image Representation and Feature Extraction” (three chapters), “Image Classification and Annotation” (four chapters), and “Image Retrieval and Presentation” (three chapters).
The first part presents the fundamental mathematical tools used in the mining of image data: the Fourier transform (chapter 1), Gabor filters (chapter 2), and the Wavelet transform (chapter 3). To understand these tools, readers should have at least a college-level understanding of integral calculus and matrix algebra. In my opinion, this section is written very succinctly, and perhaps those readers who do not have the necessary mathematical background might have to consult some other texts, for example, Digital image processing  dedicates longer and less mathematical chapters to the Fourier transform and to the Wavelet transform, respectively.
The second and third parts form the core of the book. As stated previously, the first step in the mining of image data is the extraction of features to identify the content of an image. Typically, to be understandable by human beings, these characteristics are high level. These characteristics can be concrete, like animals, churches, or boats, or abstract, like Christmas or birthdays. It turns out that extracting high-level features is not an easy task for a computer. Therefore, one way to attack this problem is to first extract low-level features, for example, identify color, textures, or geometric shapes contained in the image, and then use artificial intelligence to learn the high-level features and be able to classify the image. This is the approach followed by the author. The second part shows several methods for extracting low-level features, such as color (chapter 4), textures (chapter 5), and shapes (chapter 6). Then, the third part presents four machine learning tools: Bayesian classification (chapter 7), vector support machines (chapter 8), neural networks (chapter 9), and decision trees (chapter 10).
Finally, after studying feature extraction and image classification in Parts 2 and 3, respectively, the fourth part deals with image retrieval, that is, the process of searching in a database for images that have certain features of interest. Three image retrieval strategies are discussed: image indexing (chapter 11), image ranking (chapter 12), and image presentation (chapter 13).
The book is clearly written and the chapters follow a logical order. Almost all the figures are in color, which adds extra value to the explanation. At the end of most chapters there is a list of exercises and references in case one wants to delve further into the topic. In general, exercises refer to MATLAB functions and code. Almost all the methods presented in the book are explained using mathematical models. Thus, it is possible that more programming-oriented readers might miss the lack of pseudocode or some implementation. However, the book should be useful to anyone interested in mining image data and would certainly be a valuable addition to their personal library.