Due to growing human population and technology, huge climate change occurs which impact the environment and lives. Satellite images are profound for monitoring the ground surface. Due to the enormous availability of multispectral images, various applications based on classification, change detection have emerged. Researchers have developed lots of Multispectral Change Detection Methods (MSCD) till date. These methods show essential sub-pixel level details, such as the abundance variation of each underlying material at a given location, or the shift in material distribution within the scene, with time or as a result of significant events such as a natural disaster. The main aim of the proposed work is to present a comparative study of various unsupervised methods for detecting binary changes in multispectral imagery. Through experimental study, we provide a comparative analysis of the algorithms. The various algorithms considered for our study are (1) Principal Component Analysis (PCA), (2) PCA with K-means clustering, (3) Multivariate Alteration Detection (MAD), (4) Iteratively Reweighted Multivariate Alteration Detection (IRMAD). Using real-world multi-temporal multispectral imaging dataset, we assess and compare the performance of all these algorithms and their time efficiency on Central Processing Unit (CPU) and Graphics Processing Unit (GPU). The empirical findings, accompanied by a description of each algorithms pros and cons, are intended to help researchers pick the procedures with the good features for MSCD applications.