Research On Intelligence Image Processing Based On Blind Source Separation

With the rapid development of computer technology,digital image processing has been widely applied in remote sensing,military,biomedicine and many other fields.The currently more complex and diverse digital image are required to be analyze which impose a greater challenge to digital image processing technology.Considering the complexity of image information and the scientificity of mathematical models,the traditional optimization method shows limitation in solving the problem of digital image processing.Bionic intelligent optimization algorithm is inspired by the laws of human intelligence,biological social group or natural phenomena to imitate its various behaviors,and solve the complex optimization problems effectively.Bionic intelligent as a novel optimization algorithm shows promising application in solving digital image.Based on the framework of blind source separation,the digital image processing technology is researched by using bionic intelligent optimization.And new methodologies for image processing is developed,which summarizes as the followings:1.A differential search optimization algorithm based on the filtration guidance and random strategies was proposed.The strategy of filtration guidance was developed for initial equation search,and eliminate the disadvantageous individuals but reserve the good individuals.The updated population was more qualified and it provided convenience for updating search and also weighed the exploration capabilities and development capabilities of the algorithm in the search process.In order to further improve the optimization ability of the algorithm,the random operators were introduced to perturb the search equation in the process of population searching.Through the randomness of the perturbation method,the diversity and traversal of the algorithm could be improved,so that the algorithm could be better avoided in local convergence,and make the global converged faster.The experimental results show that the improved algorithm improves the convergence speed and accuracy.2.Under the blind source separation framework,an excellent hyperspectral unmixing algorithm was presented based on the intelligent optimization of bionics.The algorithm corporated the blind source separation technology with the hyperspectral image unmixing process by adding the abundance non-negative constraint and abundance sum-to-one constraint,and the solution were transformed into multi-objective optimization problem.And then,it was used to solve the target function using search optimization algorithm and multi-target bat optimization algorithm respectively.The experimental results show that the solution mixing algorithm based on two bionic intelligent optimization made good unmixing effects.3.Under the blind source separation framework,two excellent moving target detection algorithms were presented based on the intelligent optimization of bionics and support vector machine.The first algorithm corporated the blind source separation technology with the hyperspectral image unmixing process by adding the abundance non-negative constraint and abundance sum-to-one constraint,and the solution were transformed into multi-objective optimization problem.And then,it was used to solve the target function using search optimization algorithm and multi-target bat optimization algorithm respectively.The second algorithm exacts some fetures such as horizontal and vectical projejction,width and high ratio,and colors.The objects and video backgrounds are separated by support vector machine based on these features.The experimental results show that the first algorithm made good unmixing effects.And the second algorithm owns high target detction ratio.