The Study On The Application Of PCNN In The Color Microscopy Image Segmentation And Retrieval Of The Traditional Chinese Medicinal Materials

The informatization of Traditional Chinese Medicine(TCM)is a significant research subject which inherits and carries forward TCM.The traditional identifying method of TCM materials is simple,but it relies too much on subjective experience and knowledge,which results in high error judge rate;the physic-chemical method not only requires large-scale specialized instrument but also destroys the TCM material sample.Meanwhile,it exposes the working staff to the chemically-polluted or physically electromagnetic-disturbing environment with complex operation process and also is not good to the storage and management of identified information of TCM materials.Thus,it's an urgent task to recognize,detect,identify and retrieve the TCM materials based on modern information technology in the rapidly developing information age.From the perspective of proper processing of color micro-image of TCM materials,this thesis optimizes Pulse Coupled Neural Network(PCNN)and respectively puts forward segmentation and retrieval algorithm of color micro-image of TCM materials based on PCNN.The main content of the thesis is as follows:(1)In order to effectively extract the objective information of color micro-image of TCM materials,an improved method of 3D-PCNN automatic image segmentation is put forward.First,from the perspective of proper process of color micro-image of TCM materials,this method simplifies and improves the traditional model.Second,using the RGB color space according to the characteristic of color micro-image,it employs the maximum Shannon entropy to put different weight into the channel of image and then puts it into the improved 3D-PCNN model to be processed,and extracts the rough sketch with technology of morphology to finish the image segmentation.Finally,the segmentation process is controlled by the maximum comprehensive criterion and it is compared with maximum Shannon entropy,the minimum cross entropy and the contrast of color difference to realize the automatic segmentation of color micro-image of TCM materials.The experiment result reveals that after the segmentation,the contour and tissue of cells is clearly seen and the features and edges of different segment are enhanced.The maximum comprehensive criterion overcomes the shortcomings of maximum Shannon entropy,the minimumcross entropy and the contrast of color difference with high segmentation efficiency.Thus,the method of this thesis has better segmentation result compared with the traditional 3D-PCNN method.(2)To make the color micro-image retrieval of TCM materials more accurate and efficient,an image retrieval method based on the improved 3D-PCNN is put forward.First,according to the features of the color micro-image of TCM materials,it is processed with the improved 3D-PCNN algorithm.Second,it extracts the time sequence and entropy sequence of color micro-image of TCM materials to be retrieved.Finally,the feature extracted is matched with the feature index database and the retrieval result is returned according to the Euclidean distance with the criterion of procession ratio and recall ratio.The experiment result reveals that this method is more accurate and effective,very suitable to retrieve the color micro-image of TCM materials.