Research On Quantum Image Processing Algorithm And Its Application In Face Recognition

Classic digital image processing technology has played an important role in daily life,such as product recognition,self-service payment,vehicle recognition,face unlock,face payment,and autonomous driving.However,the increase in the number of images and the increase in resolution have brought severe challenges to the storage and calculation of classic digital image processing technologies.Quantum image processing technology combines quantum computing and classical digital image processing technology,and uses properties such as quantum superposition and quantum entanglement to improve the computational efficiency and storage capacity of classical digital image processing technology.In accordance with the order of image primary processing to image advanced processing,this paper studies quantum image spatial filtering algorithm,quantum image scaling algorithm,quantum image similarity assessment algorithm,quantum image dimensionality reduction algorithm and quantum face recognition in turn.The research content of this paper explores and expands the application of quantum computing in the field of digital image processing.On the one hand,it provides a new train of thought for the development and application of digital image processing technology.On the other hand,it shows the potential of quantum computing application scenario,and provides certain practical value and guiding significance for the improvement and promotion of the quantum computing theory.The specific research content is as follows:(1)In view of the problem that the existing quantum image spatial filtering algorithm consumes a lot of qubit resources,firstly based on the linear unitary combination method,new quantum one-dimensional linear convolution and quantum two-dimensional linear convolution are proposed,and then based on quantum two-dimensional linear convolution,the quantum image filtering algorithms are implemented,including quantum image smoothing algorithm,quantum image sharpening algorithm and quantum image edge detection algorithm.The algorithm complexity analysis shows that compared with other quantum algorithms,the quantum algorithm in this paper reduces the consumption of qubit resources.(2)Aiming at the problem that the existing quantum image scaling algorithm based on bilinear interpolation requires the image size to be exponential,this paper proposes a quantum image scaling algorithm based on the generalized quantum image representation model and bilinear interpolation method.First,use the generalized quantum image representation model to store images of any size,and then use a series of quantum gate operations with special functions to implement the quantum image scaling algorithm based on the correspondence between the color values before and after scaling given by the bilinear interpolation method.The algorithm complexity analysis shows that,compared with other quantum algorithms,the quantum algorithm in this paper allows the image size and scaling ratio to be any integer size.(3)Aiming at the problem that existing quantum image similarity measurement algorithms are limited to gray-scale images and rely on the number of quantum measurement,this paper first proposes a quantum binary image similarity assessment algorithm based on a new enhanced quantum image representation model and quantum counting method,and then uses the quantum image binarization process transforms the problem of quantum grayscale and color image similarity assessment into a quantum binary image similarity assessment problem.Algorithm complexity analysis shows that compared with other quantum algorithms,the quantum algorithm in this paper reduces the number of quantum measurements.(4)The high time complexity of the classical multi-dimensional scaling method limits its practical application.In this paper,a quantum multi-dimensional scaling algorithm based on the quantum singular value estimation method is proposed.First,use the quantum random access model and quantum linear equation solving method to calculate the quantum state corresponding to the square distance matrix,and then use the quantum singular value estimation method to obtain the quantum state corresponding to the reduced image.The algorithm complexity analysis shows that the quantum algorithm in this paper provides a polynomial speedup compared to the computational efficiency of the classical version.(5)In view of the few problems that are explored in practical application scenarios of quantum image processing technology,this paper realizes quantum face recognition based on quantum least squares support vector machine and the quantum algorithms proposed in this paper.The training process of quantum face recognition is similar to the prediction process.First use the normal arbitrary superposition state model to store the quantum image,next use the quantum image spatial filtering algorithm to suppress the noise part of the quantum image,then use the quantum multi-dimensional scaling algorithm to reduce the image dimension,and finally use the quantum least squares support vector machine to obtain the prediction model in the training process,which is used in the prediction process to get the prediction results.