In video surveillance, Person Re-Identification(Re-ID) consists in recognizing an individual who has already been observed (hence the term Re-Identification) over a network of cameras. Usually, the person Re-Id system is divided into two stages: i)constructing a person's appearance signature by extracting feature representations which should be robust against pose variations, illumination changes and occlusions and ii)Establishing the correspondence/matching between feature representations of probe and gallery by learning similarity metrics or ranking functions. A gallery is a dataset composed of images of people with known IDs whereas a probe is collected of detected persons with unknown IDs from different cameras. Specifically, the process of person Re-Identification aims essentially at matching individuals across non-overlapping cameras at different instants and locations. However, the matching is challenging due to disparities of human bodies and visual ambiguities across different cameras. This paper provides an overview of hand-crafted system for person Re-identification, including features extraction and metric learning as well as their advantages and drawbacks. The performance of some state-of-the-art person Re-ID methods on the commonly used benchmark datasets is compared and analyzed. It also provides a starting point for researchers who want to conduct novel investigations on this challenging topic.