Image category recognition is a hot topic of research in Computer Vision due to its increasing number of applications including video surveillance and medical image classification. Finding an image category is a very challenging task due to variations in illumination, scale, view point, rotation, blur, occlusion and translation. The objective of this book is to build local image features that are invariant to various geometric (e.g., changes in scale, rotation, translation), photometric transformations (e.g., illumination and blur) and improve accuracy of image category recognition, by proposing the novel image region descriptors.