A Method for New Energy Electric Vehicle Charging Hole Detection and Location Based on Machine Vision

Abstract

A new method based on machine vision is designed for electric vehicle charging hole detection and location in order to solve the low efficiency, space limitations or leakage risk in artificial charging operation for electric vehicle and to realize the automatic charging based on robot. The method enable to efficiently and accurately extract valuable characteristics of the charging hole from a charging socket image. Aim at the problem that strong electromagnetic automobile charging system will bring salt and pepper noise to image signal, the paper firstly adopts the classical median filtering for image noise cancellation. The charging socket image shows complicated background, uneven brightness, strong reflective and few goal characteristic, making the goal segmentation extremely difficult when employing the common method with fixed or adaptive threshold. Therefore, a two-stage image segmentation method based on HSI color model is proposed in paper to extract the characteristics of the charging hole target with sub-pixel precision. The image segmentation method involves threshold segmentation in the Hue component of original image, morphological operation and edge detection based on Canny operator. Meanwhile, It also reduces the influence of problem above due to the operation in Hue component. In this paper, it is based on vision platform HALCON and experiment result shows that the method enable to meet the requirements where location accuracy with sub-pixel precision and detection.