Research On Image Segmentation With Fusion Quantum-inspired And The Rate Of DNA Computing Intelligence Algorithm

Image segmentation is one of the image processing tasks from the low to high level, is a basic image processing method,and is the fundamental and key researching problem in this field, also is the main measure to analyze datum when image processed. Image segmentation technologies are related to some other fields like computer vision, machine learning, artificial intelligence, pattern recognition and so on.Researching intensively and successfully can effectively promote the development of these disciplines,and also can provide the important information when solving these complex and similar pattern recognition problems, meanwhile, the segmentation results are expected to be used in real life aspects. At present,the universal technology for segmentation doesn’t exist, generally, the concrete method has been adopted to solve the specific problem. Therefore, this dissertation has designed the quantum intelligent algorithm by analyzing the ordinary intelligent algorithm quantum theory and combining quantum theory advantages and mainly focused on the problems existing the segmentation process when just using the intelligent algorithm in the medical image segmentation. The main researching issues are following:First, it has been proved that ant colony algorithm has good robustness and good parallelism by analyzing the principles of ant colony algorithm, and it has been shown that this algorithm has weaknesses like premature,stagnation, long-time operation and inaccurate segmentation solutions by analyzing its function when applied to image segmentation.The ant colony algorithm should be quantized to solve this problem.Adopting the initial pheromone distribution formed by chaos optimization to optimize the algorithm and increase its convergence speed to solve the problem such as the low convergence speed at the initial stage and the difficulty to rapidly find a better solution when there are huge candidates, finally the algorithm flow chart used for segmentation has been given and the algorithm convergence has been verified.The effectiveness of the proposed algorithm has been verified by the standard test functions and the concrete image segmentation experiments.Second, MR images are widely used in medical clinical auxiliary diagnosis,but brain images generally includes the gray matter, brain white matter and cerebrospinal fluid and other groups, each group has complex shape structure, unclear organizational boundaries and uneven gray scale distribution that determines the the complexity division and the ordinary fuzzy clustering segmentation algorithm depending on the initial parameters set, so the effect is not good.Because of this problem, combining the fuzzy clustering algorithm and ant colony algorithm, using quantum ant search space diversity and the advantages of fast convergence speed, obtaining the optimal objective function value, setting this value as the initial parameter value to segment the image. The effectiveness of the proposed method has been verified through experimental analysis and the evaluation of objective criterion and fuzzy clustering algorithm.Third, medical CT image is the key to the follow-up treatment in the lesions and the interested region segmentation, which is one of the widely used segmentation algorithm based on the maximum entropy threshold method. But one-dimension maximum entropy threshold segmentation method is not suitable for the accurate segmentation of CT images for tumor target,2-D or higher dimensional entropy has more local spatial information of images, can be used for complex image segmentation, but each calculation process is a multiple cycle,has high computational complexity and difficult for real-time image processing system. Based on this,an improved evolutionary algorithm has been proposed by using the rich diversity of population information quantum space, the algorithm and quantum cloning operator and variations have been introduced into traditional evolutionary algorithm, searching the two-dimensional Tsallis entropy optimal threshold and giving the overall algorithm process. Experimental results show that this method is effective to solve the slow convergence speed and the evolutionary algorithm easy to fall into local extremum problem, segmentation effect is good enough to meet the requirements of medical image 3-D reconstruction.Fourth, to evaluate the segmentation algorithm not only depends on the segmentation accuracy, but also on the speed of segmentation, with image resolution becoming higher and higher, the general intelligent algorithm applied into image segmentation field is difficult to ensure real-time performance. Therefore, it is necessary to make accuracy keep high on the premise of guarantee algorithm segmentation, integrating some other new calculation methods to improve the searching the intelligent algorithm speed.Based on this, the intelligent algorithm has been proposed by introducing the biochemical reaction rate of DNA computing into the genetic algorithm, using DNA rich coding population information, introducing the adaptive crossover operator and replacement codon mutation operator into the traditional genetic algorithm, looking for the best threshold value of two-dimensional entropy, and then making the image segmented.To sum up, this dissertation has been in the view of the present problems of the intelligent algorithm applied to image segmentation, researched a medical image segmentation method based on quantum derivative intelligent algorithm and the fusion rate of DNA computing, this method has obtained better effect in precision and speed of segmentation by comparing the simulation experiment.