| At present,the research and application of color code in China is still in its infancy.Compared with the traditional two-dimensional code,color code,as a new type of identification code,can realize the functions of long-distance and multi code batch identification.Mobile phones,intelligent monitoring systems and other target detection equipment often need high-quality images and videos to achieve color code recognition.However,the brightness and contrast of the picture and video are obviously decreased under the condition of weak light,which affects the normal recognition of the color code.Moreover,the color code image is constructed based on the red,green and blue primary color blocks,resulting in the recognition of the color code image with higher requirements for brightness and contrast.Therefore,it is very important to preprocess the color code by image enhancement.Aiming at how to enhance the color code image,the following work is carried out in this paper:1.Combined with Gaussian filter and contour detection algorithm,the existing color code recognition algorithm is improved.On the basis of the research on the color code generation scheme proposed in the relevant literature,the influence of bad light on the efficiency and accuracy of color code recognition is found through experiments.2.In order to verify the effectiveness of these algorithms in different environments,we use global histogram equalization algorithm,retex based MSR image enhancement algorithm,Dajin threshold method and information entropy iterative algorithm to enhance the low gloss code image.3.In order to solve the problems found in the above experiments,an image enhancement algorithm based on convolutional neural network is proposed.The deep learning algorithm has its uniqueness in the field of low light image recognition.Compared with the traditional algorithm,it has higher model complexity to achieve the high accuracy of color code recognition.Therefore,the convolutional neural network in depth learning is applied to enhance the weak light image of color code.The main step is to use convolutional neural network to predict the illumination image of low gloss code image,and then obtain the enhanced color code image by Retinex model.The experimental results show that this method can not only enhance the color code image with different low light level,but also avoid the color distortion of the color code image. |
