| In recent years, the neural networks are widely applied to image processing,pattern recognition and other fields. It is of great significance to investigate further thecoupled neural network models and explore the principle of image processing fromthe perspective of vision bionics. Image segmentation is the fundamental step ofimage processing in computer vision, which affects the result of the subsequent imageanalysis directly. Therefore many researchers focus on finding much better imagesegmentation algorithm all the time.So far, many image segmentation algorithms have been proposed, some of themare used widely for both gray image and color image segmentation. However, furtherresearch is needed for better segmentation effect and general usage. This paperpresents a kind of color image segmentation algorithm based on harmonic whichcomes from superposition of simple harmonic waves. In order to describe the activityof neuron more appropriately, the original Kuramoto model is changed from phasecoupling to frequency coupling and globally coupling is changed to locally coupling.The instantaneous frequency is introduced to represent the phase change as aconsequence of external stimuli. The activity of the coupled neurons are reconstructedusing instantaneous frequency. The pixel values of R,G,B of color image are extractedand put into the network, three oscillating curves are produced, and they aresuperposed to produce the consonance, the new color image segmentation algorithmis formed according to the principle of synchronization of the consonance, and it isused to natural image segmentation.The performance of the segmentation are compared with other conventionalsegmentation methods, the results shows that the segmentation of the algorithm ismore smooth and accuracy. |
