Dog nose print identification method and model based on deep residual shrinkage network

Abstract

The invention relates to a dog nose print identification method based on a deep residual shrinkage network. The dog nose print identification method comprises the following steps: S10, constructing a deep residual shrinkage network model; wherein the deep residual shrinkage network model is executed by the following steps: S11: carrying out image preprocessing on an input dog nose print image to obtain an original feature vector; s12, performing feature extraction and noise filtering on the original feature vector to obtain a classification feature vector; s13, judging whether the calculation frequency of the classification feature vector reaches a preset cycle frequency or not, if not, taking the current classification feature vector as an original feature vector, returning to the step S12, and if yes, executing the step S14; s14, performing scoring prediction according to the classification feature vector, and outputting an identification result; s20, training the deep residual shrinkage network model to be convergent; and S30, inputting a dog nose print image to be identified into the trained deep residual shrinkage network model, and outputting an identification result. According to the method, redundant noise is filtered out, and the dog nose print recognition accuracy is improved.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a dog nose pattern recognition method and model based on a deep residual shrinkage network. Background technique [0002] In recent years, my country's economic level has developed rapidly, and people's living standards have also been continuously improved. At the same time, the problem of population aging is becoming more and more serious, which makes people's demand for spiritual life more intense, and the number of pet companions is also growing rapidly. Among pet companions, dogs are a good choice, but dogs are very easy to get lost accidentally due to their playful and active characters. On the other hand, the management of dogs in my country's cities is facing a severe situation, such as the increasing number of unregistered dogs and stray dogs, and the occurrence of vicious dog wounding incidents from time to time. Therefore, dog individual identification techno...

AI Technical Summary

Problems solved by technology

This detection method needs a lot of manpower and time to obtain the coordinates of the reference feature points from a large number of dog nose pattern images, and can only detect the exact coordinates of the feature points in the dog nose pattern image to be tested, and cannot classify the test image

[0005] In addition, the texture of canine nose lines is fine and the difference is small, and the very subtle noise on the nose line image will still have a great impact on the accuracy of nose line recognition. Therefore, how to improve the accuracy of dog nose line recognition is still a big problem. problem

Images