Research On Automatic Focusing Technology Based On Pulse Coupled Neural Network

Autofocus technology is widely used in modern optical imaging systems.The fundamental purpose of focusing is to keep the lens in the focused position,and the system can obtain clear images.Autofocus technology based on image processing relies on powerful digital processors,and has become a research hotspot due to its simple principle,fast response,high accuracy,and wide application range.With the continuous improvement of the computing performance of the processor,the research and application of various neural network algorithms in the field of automatic focusing technology based on image processing are becoming more and more extensive.The pulse coupled neural network is called the "third generation neural network".Its working principle is based on the information processing mechanism of the biological visual cortex,which conforms to the laws of biological vision and has a good application in the field of image processing.But in the field of automatic focusing,the pulse coupled neural network has not been studied.In this paper,an improved network model is proposed based on the traditional pulse coupled neural network.The main work of the paper includes:1.Investigating the development of autofocus technology and pulse coupled neural network respectively.The autofocus technology based on image processing is becoming more and more mature.The defocusing situation is reversed by the obtained image,and the automatic focusing is completed by combining the window construction method and focusing search strategy.The pulse coupled neural network has the characteristics of real biological neurons,and has the characteristics of pulse synchronization and capture,and is widely used in the field of image processing.Because the complex working mechanism is difficult to analyze mathematically,the research at this stage is mainly on the improvement of parameters.2.Combining pulse coupled neural networks with autofocus technology based on image processing.Proposing a new calculation method of focusing evaluation function.The image is processed by network algorithm,the grayscale difference feature of the image is extracted,and the number of ignitions experienced during network processing is recorded as the focusing evaluation function.The image segmentation algorithm is implemented on the neural network,which can segment the target and the background,realize the dynamic adjustment of the focusing window,and improve the calculation speed of the focusing evaluation function.The image enhancement algorithm is implemented on the neural network,and the network parameters are improved,so that the algorithm has a more obvious effect on the dark part of the image.3.Designing an autofocus method based on pulse coupled neural network.using a network structure to realize various image processing functions,forming a complete autofocus closed-loop system,and conceiving the execution flow of the algorithm.The simulation and experimental verification of the automatic focusing system is carried out to verify the effectiveness of the designed method,and the hardware transplantation of image data processing is initially carried out.