This thesis presents an in-depth study of the problem of photorealistic image synthesis in the field of computer graphics. To better understand the proposed solutions to this problem, we first briefly overview several relevant fundamental topics, such as probability, Monte Carlo integration, sampling methods, reflection methods, etc. We then review some of the most influential Monte-Carlo-Based image synthesis methods introduced over the past four decades. Finally, to better understand the advantages and weaknesses of each method, we evaluate the reviewed methods using the Mitsuba Renderer, an open-source, research-oriented rendering system.