[traffic sign recognition] Based on Matlab GUI YCbCr feature extraction +bp neural network traffic sign recognition [including Matlab source code 1869]

One 、BP Introduction to neural network traffic sign recognition

1 introduction (Introduction)
Traffic sign recognition is a research hotspot in the field of image processing and pattern recognition , It is widely used in driverless cars 、 Vehicle navigation and other fields , It has strong practical application value . Although the graphic structure of traffic signs is relatively simple , But to recognize traffic signs in natural environment , Facing obstruction 、 noise 、 Light variability 、 Shape distortion and other serious challenges , Therefore, traffic sign recognition is one of the key research contents in the field of pattern recognition .

2 A brief introduction to machine learning methods (Three kinds of machine learning methods)
2.1 BP neural network
BP Neural network is a multilayer feedforward network trained by error back propagation learning algorithm , Be able to learn and store a large amount of input — Output schema mapping , It is one of the most widely used neural network models .

BP The learning process of neural network consists of two processes: the forward propagation of signal and the back propagation of error . The direction of forward propagation is the input layer → Cryptic layer → Output layer , The state of each layer of neurons only affects the next layer of neurons , Get the response at the output layer , If the actual output does not match the expected output , Then enter the error back propagation stage . The error back propagation stage is back propagation from the output layer to the input layer through the hidden layer , And modify the weight layer by layer , Continuously improve the accuracy . The alternation of the two processes , Is that the network error function reaches the minimum value , So as to complete the process of information extraction and memory .

3 BP Neural network recognition of traffic signs
(1) Image preprocessing
The purpose of preprocessing is to improve the image contrast , Enhance useful information . The preprocessing operation of this experiment is gray conversion ( Adaptive threshold method ) And binarization .

(2) feature extraction
Feature extraction is to obtain the low dimensional features that can best reflect the nature of the sample by reducing the dimension . This experiment uses coarse mesh feature extraction . First, the nearest interpolation method is used to scale the sample to 70×50, Then divide the sample into 10×10 The size of the grid , Count the pixel values in each grid , formation 35 The eigenvectors of the dimensions .

(3) BP Design of neural network
BP The algorithm can be applied to feedforward networks with four or more layers , But it is easier to fall into the local minimum without convergence , Therefore, three layers are selected BP The Internet , At the same time, the training method with momentum factor is selected , Avoid falling into a local minimum .
For the choice of excitation function , Logarithm is usually used S function 、 Hyperbolic tangent S Functions and linear functions .MATLAB Three transfer functions corresponding to the above excitation functions are provided in :logsog function 、tansig Functions and linearity purelin function . In view of the experimental output of this paper , choose tansig function .
so far , The choice of the number of hidden layer nodes cannot be solved by a complete theory , However, it is necessary to make the neural network structure as compact as possible while meeting the accuracy requirements , That is to reduce the number of hidden layer nodes as much as possible . In general , The number of hidden layer nodes is determined according to the specific training results .
For the selection of training function , In this paper traingdx function , It is an adaptive learning rate method based on momentum method , Automatically adjust the learning rate , It can effectively shorten the training time .
Set the maximum number of iterations for this experiment 1000, Expect goals 1e-5, Momentum factor 0.95.

Two 、 Partial source code

function varargout = Traffic_Iden(varargin)
% TRAFFIC_IDEN MATLAB code for Traffic_Iden.fig
%      TRAFFIC_IDEN, by itself, creates a new TRAFFIC_IDEN or raises the existing
%      singleton*.
%
%      H = TRAFFIC_IDEN returns the handle to a new TRAFFIC_IDEN or the handle to
%      the existing singleton*.
%
%      TRAFFIC_IDEN('CALLBACK',hObject,eventData,handles,...) calls the local
%      function named CALLBACK in TRAFFIC_IDEN.M with the given input arguments.
%
%      TRAFFIC_IDEN('Property','Value',...) creates a new TRAFFIC_IDEN or raises the
%      existing singleton*.  Starting from the left, property value pairs are
%      applied to the GUI before Traffic_Iden_OpeningFcn gets called.  An
%      unrecognized property name or invalid value makes property application
%      stop.  All inputs are passed to Traffic_Iden_OpeningFcn via varargin.
%
%      *See GUI Options on GUIDE's Tools menu.  Choose "GUI allows only one
%      instance to run (singleton)".
%
% See also: GUIDE, GUIDATA, GUIHANDLES

% Edit the above text to modify the response to help Traffic_Iden
% Begin initialization code - DO NOT EDIT
gui_Singleton = 1;
gui_State = struct('gui_Name',       mfilename, ...
    'gui_Singleton',  gui_Singleton, ...
    'gui_OpeningFcn', @Traffic_Iden_OpeningFcn, ...
    'gui_OutputFcn',  @Traffic_Iden_OutputFcn, ...
    'gui_LayoutFcn',  [] , ...
    'gui_Callback',   []);
if nargin && ischar(varargin{
    1})
    gui_State.gui_Callback = str2func(varargin{
    1});
end

if nargout
    [varargout{
    1:nargout}] = gui_mainfcn(gui_State, varargin{
    :});
else
    gui_mainfcn(gui_State, varargin{
    :});
    
end
% End initialization code - DO NOT EDIT

3、 ... and 、 Running results

Four 、matlab Edition and references

1 matlab edition
2014a

2 reference
[1] Cai Limei .MATLAB The image processing —— theory 、 Algorithm and example analysis [M]. tsinghua university press ,2020.
[2] Yang Dan , Zhao Haibin , Long Zhe .MATLAB Detailed explanation of image processing examples [M]. tsinghua university press ,2013.
[3] Zhou pin .MATLAB Image processing and graphical user interface design [M]. tsinghua university press ,2013.
[4] Liu Chenglong . Master MATLAB The image processing [M]. tsinghua university press ,2015.
[5] Zhang Xingguo , Liu Xiaolei , Li Jing , Wang Huandong .BP Real time detection and recognition of speed limit traffic signs based on neural network [J]. Journal of Xi'an University of Electronic Science and Technology . 2018,45(05)
[6] Zhong Ling , Yu Yajie , Zhangzhijia , Jinyongchao . Comparison and analysis of traffic sign recognition algorithms [J]. Software Engineering . 2016,19(01)