Generic method for Error Estimation and Modeling in Depth Data Using an Embedded System
Stereo Vision, Depth error, Sterolabs ZED, NVIDIA Jetson TK1, OpenCV, Cloud Point
In this work we present the advances in the proposal of a generic method for estimating the error in the depth data provided by any device capable of generating a color image and a disparity map at the same time. In addition, the basis for obtaining a mathematical model of the error in the depth data provided by the ZED camera is established to determine how accurate it really is. An embedded system based on the NVIDIA Jetson TK1 board is implemented to prove the error estimation method in the ZED camera. This will be the basis for future work such as obtaining a real-time 3D reconstruction as accurately as possible using various point cloud capture devices and GPUs, and in general for any application that needs as a useful parameter the accuracy of the data from Depth provided by the ZED camera.