This paper deals with vision-based localization techniques in structured underwater environments. For underwater robots, accurate localization is necessary to perform complex missions successfully, but few sensors are available for accurate localization in the underwater environment. Among the available sensors, cameras are very useful for performing short-range tasks despite harsh underwater conditions including low visibility, noise, and large areas of featureless scene. To mitigate these problems, we design artificial landmarks to be utilized with a camera for localization, and propose a novel vision-based object detection technique and apply it to the Monte Carlo localization (MCL) algorithm, amap-based localization technique. In the image processing step, a novel correlation coefficient using a weighted sum, multiple-template-based object selection, and color-based image segmentation methods are proposed to improve the conventional approach. In the localization step, to apply the landmark detection results to MCL, dead-reckoning information and landmark detection results are used for prediction and update phases, respectively. The performance of the proposed technique is evaluated by experiments with an underwater robot platform and the results are discussed.