Container number automatic acquisition and identification method in railway container loading and unloading process

Abstract

The invention discloses a container number automatic acquisition and identification method in a railway container loading and unloading process, which comprises the following steps of: respectively acquiring images of the top surface and the side surface of a container, and carrying out general preprocessing on the images of the containers with different container body colors; judging the printingcolor of the container number on the basis of image binaryzation; analyzing and positioning a printing area of the box number through a connected domain; judging the arrangement mode of the containernumbers according to the length-width ratio threshold of the minimum external moment of the container number area, and respectively carrying out character segmentation in different direction projection sequences; building three neural network model classifiers to recognize three parts of characters in the container number, and mutually verifying container number recognition results of the top face and the side faces of the container. According to the method, the influence of local character stripping caused by abrasion on an identification result is overcome, container numbers of different colors and arrangement modes are identified, and rapid and accurate identification of the container numbers is completed in combination with mutual inspection of the container numbers of different surfaces and self-inspection of container check codes.

Description

technical field [0001] The invention relates to the technical field of image intelligent recognition, in particular to a method for automatic collection and recognition of container numbers during loading and unloading of railway containers. Background technique [0002] As a container, such as figure 1 As shown, it is widely used in the global logistics and transportation industry. Containers not only have a variety of specifications and sizes, but also the color of the surface of the box is not unique. In order to realize the global unified management of containers, it is stipulated that each container put into operation must print a unique container number, such as figure 1 shown in the white box. The container number consists of 11 characters, which can be divided into 3 parts: ① 4 uppercase English letters; ② 6 Arabic numerals; ③ 1 digit, which is the check code. Railway container yards have strict regulations on the management of each container, including unloading...

AI Technical Summary

Problems solved by technology

[0005]1) The collection data of the existing box number is single

The existing method is to install a camera on the lock base of the crane to take pictures of one side of the container. Due to the wear and tear of the container, the characters peel off, so it is difficult to ensure the integrity of the container number when collecting images of one side

[0006]2) Poor versatility in identifying box numbers in white and black fonts

First, the box numbers in white font and black font represent the background and target respectively after binarization, and it is difficult to distinguish the box numbers of different font colors with a fixed threshold

[0007]3) The identification result of the box number lacks verification

The existing method is mainly to extract the container number from a single container picture on a specific surface, and does not distinguish between capital letters and numbers in the container number. The recognition result of the container number only depends on the information of a picture, and there is no Cross-validation using box number recognition results from different faces

Images