Research On Quantum Image Matching And Classification Algorithms

In recent years,quantum computing has received extensive attention due to its outstanding computing power,which has attracted many researchers to join the field of quantum computing.Due to the wide application of image processing,quantum image processing combined with quantum computing has become an important front for quantum computing research and has great practical significance.Most of the existing quantum image processing algorithms do not take into account the measurement problem.If users want to a results,they must measure the final state many times to get the value of all pixels' values.In addition,the algorithm can only measure one final state at a time.To measure many times,users must execute the algorithm many times.Therefore,the effectiveness of the algorithm is open to question.Quantum image matching is an important branch of quantum image processing,but also the basis algorithm of image vision,image understanding.Existing quantum image matching schemes also have measurement problems.In this paper,for simple image matching problems,we propose a quantum image matching algorithm with effective measurement.This scheme allows the target pixel to be measured with a higher probability by adjusting the probability of each pixel.Complexity analysis shows that the algorithm has only linear complexity.Considering that the pixel values in the image are not unique,we improve and propose a stable quantum image matching algorithm to ensure the correct output of the unique coordinates in the complex image matching problem.Complexity analysis shows that the algorithm still has an efficiency advantage over the classical algorithm while correctly handling complex image matching problems.Another important branch of quantum image processing field is quantum image classification.In recent years,the image classification scheme based on machine learning algorithm improves the classification accuracy to a new height.Because of the high complexity of many machine learning algorithms,these schemes are unavoidable to be inefficient.This paper presents an efficient classification algorithm for classical image,namely image classification algorithm based on quantum KNN.Quantum KNN algorithm improves the overall efficiency of the image classification by parallelizing the distance calculation and reducing the complexity of the search process.The complexity analysis and simulation experiment show that the scheme can improve the efficiency while ensuring the good accuracy.