Stereo Image Matching Method Based On Hierarchical Algorithm

Stereo matching is an important research direction in the theory and application of computer vision.At present,image matching algorithms are divided into two categories:local optimal matching algorithm and global optimal matching algorithm.Local optimal matching algorithm mainly relies on some physical features and information of the image in the matching process.This kind of algorithm has lower complexity and less computational complexity.The performance of global optimal matching algorithm is much higher than that of the former,and the matching results are more accurate,which can basically meet the application requirements,but the matching complexity may be relatively higher.With the deepening of research,some mathematical optimization methods have been introduced,and genetic algorithm is one of them.The algorithm has high robustness.The application of genetic algorithm to stereo image matching algorithm can improve the matching accuracy of the algorithm.This paper presents a hierarchical matching algorithm based on improved genetic algorithm.The hierarchical algorithm improves and fuses the traditional matching algorithm with the function optimization method.Firstly,the traditional genetic algorithm is improved to improve the matching accuracy and enhance the matching performance of the algorithm.Secondly,the traditional matching algorithm is combined with the improved genetic algorithm to achieve double optimization of speed and accuracy.In the first layer of the hierarchical algorithm,the improved genetic algorithm is adopted: the selection operator adopts the best retention strategy,which effectively reduces the mismatch rate and improves the accuracy of the algorithm;the crossover operator uses the self-recognition crossover operator to avoid local convergence and improve the global performance of the algorithm.The disparity of the image feature points is obtained by the improved genetic algorithm.On the basis of the first layer,the feature points of the original image are removed.The remaining points are collectively called non-feature points.The disparity of non-feature points is obtained by using region matching algorithm to match non-feature points as the second layer.In the second layer,the region matching algorithm is used to improve the matching efficiencyof the algorithm.In addition,the hierarchical algorithm also takes into account the noise problem and makes corresponding processing.By using Canny operator with higher anti-interference as feature extraction operator,Canny operator is less sensitive to noise than other operators.Canny operator smoothes the image in the process of feature extraction and optimizes the extracted edge.The hierarchical algorithm proposed in this paper can combine the advantages of both algorithms into the hierarchical algorithm.The experimental results are compared by changing the threshold of the hierarchical algorithm.The experimental results show that the change of some parameters in the hierarchical algorithm may affect the matching results.For different matching graphs,different matching thresholds are used to compare,and the corresponding matching results are obtained.