A Micro-Vascular Image Segmentation Method Based on the Improved Adaptive Region Growing

摘要

Due to the characteristic of the blood flow in capillaries, the color distribution of the Micro-vascular image is uneven and its gray scale continuity is poor. At the same time, Micro-vascular image also has the characteristic that light color, low contrast with background and so on. Therefore we put forward a Micro-vascular image segmentation method based on the improved adaptive region growing. Firstly this method uses homomorphic filter to adjust the brightness of the image, enhance the visibility of the micro-vascular and the background, and make it clearer. Then use the average gray level of local area and the local average gradient instead of global average gray level and gradient. Thus, we got the adaptive region growing algorithm which improved by the traditional region growing method. Finally we based on the selected seed to split the capillaries. This algorithm is mainly on the characteristics of the capillary images, therefore, compared with other traditional image processing algorithm has better segmentation effect.