Visual-SLAM 키워드 리스트

Posted on Edited on In , , Views:

Camera Model + Projective Geometry

  • Digital Image의 특성
    • 해상도, 컬러 채널, 대비
    • Point, Line, Edge, Blob
  • 카메라 / 렌즈 하드웨어
    • Sensor Resolution, Aperture, ISO, Exposure, Shutter Speed
    • Noise model, convolution, filtering, Fourier analysis
  • 영상처리
    • rgb2gray, image resize, template matching
  • Pinhole 카메라 모델, 왜곡, 캘리브레이션
    • Pinhole 카메라 모델의 영상 투영
    • Radial / Tangential 왜곡 모델
    • Zhang 카메라 캘리브레이션
  • Projective Geometry
    • Vanishing point
    • Euclidean / Homogeneous coordinates
    • Euler Angle, Axis-angle, Quaternion, SO(3), SE(3)
    • Coordinate transformation

Feature Detection / Descriptor / Matching

  • Feature Detector
    • Properties of corner, Scale space
    • Harris / Shi-Tomasi
    • SIFT / SURF
    • FAST / oFAST (ORB)
  • Feature Descriptor
    • SIFT
    • BRIEF / rBRIEF (ORB)
  • Deep local feature
    • SuperPoint
    • D2Net
  • Feature Matching
    • Brute-Force matching
    • FLANN - Nearest Neighbour
    • LSH / Multi-probe LSH
    • HBST
  • Optical flow, KLT Tracker
    • Horn & Schunck L2 / LK / L1 Reg Optical flow
    • KLT Tracker
  • Global feature
    • Bag-of-visual-words
    • VLAD
    • NetVLAD

Multiview geometry

  • Epipolar geometry (2D-2D)
    • Essential / Fundamental Matrix (5/8-point)
    • Singular Value Decomposition (SVD)
    • Homography matrix (4-point)
  • Multiple view (2D-3D)
    • Perspective-n-Points Problem (PnP)
    • Triangulation / Disparity
  • Transformation between point clouds (3D-3D)
    • Iterative Closet Points Problem (ICP)
  • Outlier Rejection
    • RANSAC
    • Robust Estimator / Maximum consensus problem
    • Convex relaxation
  • Bundle adjustment / Structure from Motion
    • Numerical Methods and optimisation
    • Least squares problem
    • Gauss-Newton, Levenberg-Marquardt algorithm
    • Non-linear solver libraries / factor graphs
    • Motion-only bundle adjustment / Full BA

Simultaneous Localization and Mapping

  • Introduction to SLAM
    • Definition of SLAM
    • Difference between SLAM, SfM, Odometry, Path planning, Image stitching
    • Indirect (feature-based) vs Direct-based SLAM
  • Loop closure / Bag of Visual Words
    • Descriptor database formulation / query
    • kd-tree
  • Mapping
    • Stereo matching
    • OctoMap
    • TSDF
    • Voxel Hashing
  • Sensor fusion / Tracking
    • KF, EKF, UKF
    • Factor graph optimisation
    • VINS estimator
  • Rolling Shutter compensation

Machine Learning

  • Recognition
    • HOG, SIFT, GIST
    • Viola-Jones, Cascades
    • AdaBoost, kNN, SVM, Random Forest
  • Principle component analysis
    • PCA, ICA, CCA
  • Estimation
    • EM Algorithm → ML, MAP
  • Clustering
    • K-means, Mean-shift, Pedro-clustering

Artificial Neural Network

  • Neural Network
    • Backpropagation, Multi-layer perceptron, SoftMax
    • Optimization, Stochastic Gradient Descent SGD), ADAM
    • Batch Norm, Dropout, Momentum
    • Ensemble, Transfer Learning, Data augmentation
  • Convolutional Neural Network
    • Convolution, Pooling
    • AlexNet, VGG, GoogLeNet, ResNet, ResNeXT, EfficientNet
    • Sparse Convolutional Neural Network
  • Recurrent Neural Network
    • RNN, LSTM, GRU, Attention, Transformer
    • Soft Attention
  • Detection
    • YOLO, MobileNet-SSD, RCNN,
  • Segmentation
    • U-Net
  • Pose Estimation