Medical Image Enhancement Based On Improved PCNN Image Factorization

Medical imaging technolgy is an important assisting means to clinical diagnosis and image enhancement is of great importance in medical image processing.Although many image enhancement methods have already been used,researchers still keep exploring an efficient and fast one in the field.Pusle coupled neural network (PCNN),the model of which is first put forward in the last 90's decade,is called the third generation artifitial neural network.lt has been widely used in various applications in image processing.This thesis tries using an novel image enhancement method based on human visual properties and improved PCNN image factorization to enhance medical images.The paper concentrats on the following research points:This paper decribes the PCNN image factorization system proposed by Johnson and analyzes the theory of the image factorization system as well as the vital parameters affecting the factorization procedure.Besides,point out the flaws of the original PCNN image factorization system.For the shortcomings of the original PCNN image factorization system, an improved PCNN image factorization system is put forward.The major improvements include two parts:first, get rid of the PCNN#1 layer in the original system and second, different image factors utilize non-equilibrium linear-threshold decay processes. The improved system greatly speeds up the factorization procedure and also controls the layers of the image factors.Using the improved PCNN image factorization system and some results from human visual properties, the paper proposes a medical image enhancement algorithm based on the improved PCNN image factorization and human visual properties. The result is compared with the ones using common image enhancement algorithms.Finally,analyze some weaknesses in this new algorithm and and point out the next steps to improve the idea.