Automatic Segmentation Algorithm Study On 2D-MAP

2D-MAP image processing technology is the use of image segmentation to process the map of theγphoton detector crystal space location inorder to identify scintillation incident location, and generate crystal position map (CPM). The article uses the following methods of 2D-MAP image segmentation. use the time and frequency domain filtering image noise reduction pretreatment method to improve the effect of the watershed 2D-MAP image segmentation algorithm; based on 2D-MAP physical distribution model characteristics, mathematical morphology tophat and bothat to improve the watersh transformation segmentation; use the self-organizing feature map neural network to train the 2D-MAP pixel sample data of the image, through the network weights vector space distribution features to represent the geometric distribution structure, and cluster the 2D-MAP of crystal response position; use the PNN probabilistic neural network to identify the location of each crystal scintillation incident and determine crystal location (CPM) by the largest posterior probability criteria in the Gaussian probability distribution model; use the image search algorithm to identify 2D-MAP independent scintillation incident highlight dot regional distribution, and through regional gray crystal maximum determine the crystal detectors centres, and through the average distance from the centres to identify crystal border image segmentation.PET data transmission system uses the EPP parallel port protocol to send parallel data, through the time-sequence handshaking signal to control the receiving of the parallel data, and transforms it into the serial format including the address and data information to meet the need of mass data and high speed transmission.