Speckle dots have the advantage of easy projection, which makes them good candidate features of structured light (SL) cameras, such as Kinect v1. However, they generally yield poor accuracy due to block matching. To improve their accuracy, this paper proposes a dot-coded SL, the coding information of which is added into dot distribution. Some of the dots are arranged regularly to provide easy-to-locate corner features, while others are specially designed to form different shapes of unique identification. A Gaussian-cross module and a simplified ResNet have been proposed to conduct robust decoding. Various experiments are performed to verify the accuracy and robustness of our framework.