Method and device for improving serial number recognition rate based on multi-channel synthesis technology

Abstract

The invention discloses a method and device for improving the serial number recognition rate based on a multi-channel synthesis technology. The method comprises the steps: collecting images of a to-be-identified banknote under different spectrums, and selecting two or more types of spectral imaging images according to the difference of all spectral imaging features; for each selected spectral imaging image, positioning and extracting a corresponding serial number area to obtain a serial number area image of each spectral imaging image; carrying out image synthesis on the positioned and extracted serial number area images of the spectral imaging images; and performing serial number identification on the synthesized image to identify serial number characters in the synthesized image. According to the method, different channel combinations are adopted to carry out serial number image synthesis and then segmentation and identification, so that the resolution is improved, detail features are reserved, and the identification rate is improved; and the method can be widely applied to financial machine and tool products needing serial number identification, such as money counters and sorters, and the serial number identification efficiency is greatly improved.

Description

technical field [0001] The invention relates to the field of serial number recognition, in particular to a method and device for improving the recognition rate of serial numbers based on multi-channel synthesis technology. Background technique [0002] The current financial equipment products have very high requirements for the recognition rate of RMB banknote serial numbers, and the general misrecognition rate is required to be within 3 / 10,000. The acquisition time is extremely short. When collecting images of banknotes, the machine needs to collect image data under multiple spectra at one time according to the requirements of identification and authentication, which means that the lighting time and image acquisition time allocated to each spectrum are greatly compressed. As a result, the vertical resolution of the image collected under each spectrum is low. Under the limited image height, the serial number area will be compressed accordingly, and the corresponding charact...

AI Technical Summary

Problems solved by technology

[0004] The existing method of selecting a single visible light channel for identification generally selects the green light channel (because the serial number area under the green light channel is complete) for positioning, segmentation and identification, but is limited by the relatively low vertical resolution of a single channel. Low, the detail features are not fully preserved, resulting in unsatisfactory recognition rate

Images