Image Matching Based On Improved Particle Swarm Optimization Algorithm

With the development of science and technology,widely used matching technique has become an important field in computer vision and image processing.But new problems are also emerging with the expansion of this application as one of the important research issues lies on how to improve the efficiency of image matching.Two predominant aspects are taken into account for this topic: on the one hand,the research could be focused on the new formula of similarity measurement;on the other hand,attention can also be paid to the optimal image matching search algorithm.However,many practical problems in reality can be transformed into an optimization problem,such as knapsack problem,the problem of path planning,WSN routing protocol etc.Opti mization problem has been widely used in various fields of social science,which can effectively solve the problem of scientific allocation of resources,production planning,urban planning and other issues.In recent years,swarm intelligence algorithm has become a research hotspot.PSO algorithm as one of the swarm intelligence algorithm has been increasingly applied to the field of image matching,but this algorithm is prone to premature convergence.Consequently,the improvement of PSO algorithm has als o become a hot research issue.Firstly,in order to improve the fact that the PSO algorithm is haunted by problems such as local optimum,long searching time and low matching accuracy,solution that improving and adding interference into the control factor ? in particle velocity formula is proposed.First of all,control factor ? is improved in form of logarithm and the improved nonlinear inertia weight is used to optimize the PSO algorithm,which could help balance the particles' global and local search ability in solution space.When?is relatively larger,a larger ? can make the algorithm unconfined in small area and jump out of this area to reach the previously unsearched region.When ?is relatively smaller,the smaller ?can strength algorithm search ability in small area,which makes the algorithm search in the vicinity and speeds up the convergence degree of algorithm.What is more,adding dynamic disturbance on velocity update formula can disturb the speed,which makes the improved algorithm avoid the dilemma that most particle velocity stagnates to zero and stops the search so as the algorithm falls into premature convergence in the latter stage of the algorithm when particle doesn't find the optimum solution.Secondly,according to the drawbacks of PSO algorithm in image matching,this paper also proposes the improved Metropolis criterion and introduces it into the improved particle swarm algorithm,on the basis of which the campaign thought is put forward.Applying the improvements mentioned above in image matching,the algorithm can make the particle jump out of optimal solution in local loop area and get into the global area of algorithm,which makes it closer to the best solution.In order to verify the validity of the two algorithms,this paper uses MATLAB to simulate and verify the proposed algorithm and compares them with the standard PSO image matching algorithm.The results show that the proposed PSO algorithm in this paper improves the efficiency and robustness of image matching to a great extent.This paper use three types of performance index to measure the quality of image matching: average running time of image matching,anti-noise performance and the matching accuracy.In order to fully explain the effectiveness and feasibility of the proposed image matching algorithm,this paper compares relationship curve between the number of iterations and adaptation value and the robustness of map matching between no-noise and Gaussian noise of proposed algorithm and PSO image matching algorithm.The simulation results show that the proposed image matching algorithm is better than POS in the search speed,robustness and the ability to jump out of local optimal solution,which demonstrates the effectiveness and feasibility of the proposed algorithm.