**Abstract**

Burn injuries are the fourth leading cause of death among accidental deaths, which is the third leading cause of death in humans. Early treatments are required to stop the degeneration of the muscles, which could lead to the complications of the patient, even death. A proper and in time, treatment can lead to a full recovery of the person. Professionals often mislead these early treatments due to the stress and the lack of sleep during their job. Thus, an artificial medical assisting method was developed to detect and classify burns. Skin burns in color images were accurately detected and classified according to burn degree to assist clinicians during diagnosis and early treatment. Detection and classification of burnt areas are performed using the sparse representation of feature vectors with over-redundant dictionaries.

**Keywords:** burn injuries, sparse representation, over-redundant dictionaries, burn classification and detection
