An Improved Dog Face Detection And Recognition Algorithm

The continuous prosperity and development of the pet dog industry have increasingly affected the national economy and people's lives.As a result,the demand for management of pet dogs,the recording of basic information for dogs,dog health insurance and pet feeding have increased.For this reason,dog face detection and dog face recognition technology based on machine learning came into being.Different from traditional ID tags and artificial chip injection methods,there are many advantages in efficiency,reliability,cost and operability.Therefore,based on the related improvements of facial recognition technology,this paper designs a set of dog face detection and dog face recognition system.At the same time,it introduces the commonly used face detection methods and the commonly used facial recognition technology algorithms,and compares the experiments with the algorithm.The main work is as follows:In the process of extracting features in traditional HOG,the computational complexity of the algorithm is high,which is not conducive to feature extraction,so the detection process is rather lengthy and the detection rate is not high.It is proposed that the convolutional layer of the first layer of the convolutional neural network is used to extract the features of the edge,shape and color of the image,which saves the time of HOG feature extraction,improves the efficiency and the accuracy of detection rate.In the traditional 2DPCA dog face recognition process,the limitation of feature extraction of ranks and columns is only linearly dimension-reduced in the direction of rows and columns.The feature extraction of multi-angle 2DPCA is proposed.Under the condition of ensuring accuracy,firstly the self-correction of the tilt angle is performed,and then the feature extraction of 2DPCA is performed at an angular interval,which breaks the limitation of the conventional algorithm,and can more accurately extract the dog face features and the higher recognition rate.