Analyzing medical images involves challenging tasks because of the minute variations and larger data set for analysis. Medical imaging establishes a database of normal anatomy and physiology to identify the abnormalities in the dataset. Thermal imaging is widely accepted as an accurate and reliable tool for medical assessment and diagnosis. It is used in a variety of medical applications in the fields of neurology, oncology, orthopedics and dermatology. It also allows for screening a huge number of people within a very short time frame. Image classification is an important and tedious task in various applications domains, involving biometry, biomedical imaging, remote sensing, etc. The input for a classification is a set of training records where each record has several attributes. Due to medical reasons, there exists challenges in image registration resulting from image noise, lack of sufficient number of images, resolution difference between images. This paper surveys about various image registration, feature extraction, feature selection and image classification techniques in order to identify the optimal techniques to classify the normal and abnormal images. Image registration aligns the images from different datasets. Moreover, the comparison between various image registration, feature extraction, selection and classification techniques are depicted. © Research India Publications.