What is the difference between “3D reconstruction of RGB-D cameras” and traditional SFM and SLAM algorithms?

First, the input data stream is different. In the 3D reconstruction based on RGB-D camera, depth information and color information are input, and depth information plays a leading role, while in SLAM and SFM algorithms, the input is mainly color information, and individual works also combine depth information.
Second, the focus is different. In the SLAM framework algorithm, the positioning is still the main body. We usually need the positioning to respond in real time, and the construction of dense maps is usually large in scale and computational complexity, so the construction of maps does not necessarily require real time. Of course, when the map is relatively sparse, it can also be constructed in real time, but it does not meet the actual needs. Another idea is to build a map as the main body, followed by positioning, aiming to build a dense and accurate high-quality map, this high-quality map can be provided for the SLAM algorithm for positioning, and the three-dimensional reconstruction based on RGB-D cameras” is One way to meet the above needs.