Power cable detection method and device based on convolutional neural network and frequency domain enhancement

The invention discloses a power cable detection method and device based on a convolutional neural network and frequency domain enhancement, and the method comprises the steps: carrying out the preprocessing of an original image, manually marking a wire region in an original image containing a wire,...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: FAN XINLIN
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention discloses a power cable detection method and device based on a convolutional neural network and frequency domain enhancement, and the method comprises the steps: carrying out the preprocessing of an original image, manually marking a wire region in an original image containing a wire, obtaining a marked wire region, carrying out the wavelet transformation of the original image, and obtaining an original image; obtaining a low-frequency sub-image and a high-frequency sub-image; a network model building step: building a network model comprising a coding network and a decoding network; a network model training step: obtaining an error of a prediction result through the predicted wire area and the marked wire area, correcting the network model according to the error, and repeatedly training to obtain a trained network model; and a power cable detection (prediction) step: carrying out power cable detection (prediction) by using the trained network model. According to the power cable detection method