Zhanhao Wu
China Mobile Hangzhou
R&D Center
  • Email
  • wuzhanhao@cmhi.
    chinamobile.com
Recent Research Direction
Smart Review, Album Classification
Skills
Familiar with C++, Python language, familiar with Tensorflow, OpenCV library functions. Familiar with camera calibration, 3D reconstruction, SLAM and other visual aspects and Deep Learning knowledge of CNN networks.
Education Experience
2016-2019 Donghua University, Master Student
2012-2016 Donghua University, Bachelor

1

Projects
Grasping system based on depth camera and robotic arm (Beijing Inspire-Robot internship)
Project Description: Use the depth camera to acquire the scene point cloud, and then use the point cloud matching algorithm to obtain the pose of the target object and the camera, and then the result of the hand eye calibration, control the robot arm to grab the target and transfer to other targets. Exhibition at the World Robotics Competition held in Beijing in August.
URL: https://v.qq.com/x/page/b0762wy2oe8.html
Main task: calibration of camera and robot arm (including hand-eye calibration), dual-target setting of depth camera, control of arm movement and point cloud acquisition, target point cloud matching to obtain pose
Mechanical arm guessing system based on CNN and LSTM
Project Description: Using the images taken by the RGB camera, extract features using VGG-16, and put them into LSTM for classification. Mainly responsible for: data acquisition and marking, design and implementation of objective function algorithm
3D reconstruction system based on multiple RGB-D cameras (personal implementation)
Project Description: Using multiple spatial dual-infrared cameras to perform fast 3D reconstruction of objects in a certain space, initially applied to real-time measurement of volume.
Main task:
Design and Implementation of Fast Calculation and Point Cloud Adjacent Retrieval Algorithm for Rough Bound Box in Space Cutting
Multi-camera external parameter calibration algorithm's implement, which mainly uses C++, OpenCV, and ceres to implement a global optimization algorithm.
Depth camera line structure optical ranging (Shanghai Percipio Technology project)
Project Description: Calibration of depth cameras and laser emitters using robotic arms.
Main task:
Hand-eye calibration - get the external reference of the robot arm and the checkerboard
Calibration of depth camera and laser transmitter
Light projection - using known internal and external parameters, making a reverse projection according to the focal length, intersecting the plane of light formed by the laser emitter to obtain depth.
Papers & Patents
A Versatile Homography Computation Method Based on Two Real Points.
Image and Vision Computing.(SCI, IF:2.67).
Third author, propose a general two-point decomposition formula based on two real points.
Camera Calibration with Coplanar Conics: A Unified Explanation and Ambiguity Analysis.
IPSJ Transactions on Computer Vision and Applications.
Second author, A novel method for calibration using coplanar conic curves.
Patent: A high-precision multi-feature planar template and its distortion optimization and calibration method