Vehicle matting method, system and device based on neural network and storage medium

• Main Authors: ZOU YU, HU HONG, CHENG DANNI, LUO CHAO, SHAO MINHAO
• Format: Patent
• Language: Chinese; English
• Published: 30.04.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention provides a vehicle matting method, system and device based on a neural network and a storage medium. The method comprises the steps: inputting a to-be-recognized image into a trained first convolutional neural network, and extracting the multi-feature information of each pixel point in image information, wherein the multi-feature information at least comprises color information and object recognition information of each pixel point; when the object recognition information is the pixel point of the vehicle, morphological operation is carried out on the area where the object recognition information is the pixel set of the vehicle, and a to-be-recognized area is obtained from the area where the edge of the area is corroded and expanded; inputting the multi-feature information of the pixel points in the to-be-recognized area into a trained second convolutional neural network to at least output a to-be-matted track of a vehicle; and performing matting on a closed graph region surrounded by the to-be-