Injurious or Noninjurious Defect Identification From MFL Images in Pipeline Inspection Using Convolutional Neural Network.
Jian FengFangming LiSenxiang LuJinhai LiuDazhong MaPublished in: IEEE Trans. Instrum. Meas. (2017)
Keyphrases
- image data
- ground truth
- three dimensional
- convolutional neural network
- image analysis
- edge detection
- image database
- lighting conditions
- visual inspection
- image classification
- image registration
- image features
- object recognition
- satellite images
- test images
- input image
- processing pipeline
- image collections
- defect detection
- identification rate
- illumination conditions
- image retrieval
- region of interest
- image matching
- spatial information
- similarity measure
- quality control
- rigid body
- keypoints
- multi class
- multiscale
- feature selection