Key elements extraction based on particle feature and RBFNN in new meter calibration method

Imitating human's ability of visual and analysis models, a computer vision-based measuring-instrument calibration method is proposed in this paper. Capturing meter images from real-time video with every analog output, via radial basis function neural network (RBFNN) to classify the dial plate elements, and meter reading can be obtained from complex needle movement. The particle feature extraction, RBFNN simple architecture and fast training process can satisfy the real-time meter reading perfectly in the needle identifier and reading recognition. The least squares fit method based on maximum likelihood estimate is efficient in needle parameter derivation, and the computer vision-based meter calibration method has high performance and accuracy