In recent years, underwater robotics vehicles (URVs), including remotely operated vehicles (ROVs) and autonomous underwater vehicles (AUVs), are growing rapidly in deployment for both commercial and military applications. Among various imaging approaches used in these vehicles, optical vision has shown great advantages in intuitive representation and extremely high resolution, when compared with a popular alternative like acoustic imaging. Unfortunately, optical images are severely degraded due to the absorption and scattering of light in turbid water environments. Performance of a conventional camera system directly depends on the turbidity of water. Green laser has been used as an active light source because of its good transmittance in water and its high output power. At present, novel lighting and imaging systems, such as laser galvanometer scanning systems and range-gated systems have shown some achievements in minimizing light scattering effects. However, their unwieldy sizes and high power consumptions limit their usages on small URVs. This thesis discusses a low-powered and low-speed scanning scheme, which has an adjustable illumination volume to fit various water turbidities. The illumination beam can be concentrated to improve image signal-to-noise ratio in turbid water, and expanded to improve frame update rate in clear water. The scheme has advantages in simple structure, compact size, and low power consumption. An illumination model is developed to analyze the relationship between light flux, illumination volume, and water turbidity. It focuses on highly turbid water condition, and assumes that forward and backward scattered fluxes uniformly fill all forward and backward directions. Light propagation is divided into two stages: from light source to object, and from object to camera. The light flux in each stage is regarded as a linear superposition of transmitted, forward-scattered, and backscattered components, which are individually calculated with volume scattering function and Bouguer's law. The model differs from existing models as it considers geometric parameters in an illumination scene. It quantitatively reveals how illumination volume affects the distributions of flux components. When a light beam is being expanded, the total flux received by a camera increases, and backscattered component increases faster than the correctly transmitted (or desired) component. When a wide illumination beam is used in turbid water, the signal strength of the scattered noise may exceed that of the desired signal component. This makes the objects undetectable.