A data sampling method of 3-D ultrasound images based on improved octree decomposition

In the field of medical image processing, 3-D(three dimensional)ultrasound image data is characterized by large amount and heterogeneity, which makes the image data processing more complicated and time-consuming. Therefore, data sampling is necessary to achieve the high efficiency and fluency of mass data processing. This paper presents a data sampling method of 3-D ultrasound images based on improved octree decomposition, which can obtain representative data of 3-D ultrasound images automatically and efficiently. Firstly, the segmentation threshold can be determined by using image segmentation algorithm based on fuzzy set theory. Then the improved octree decomposition algorithm is applied to process 3-D ultrasound images.Finally, according to the selection criteria, an optimal homogeneous cube and a typical heterogeneous cube are selected as sampling results to output. This method takes fully consideration on the spatial relationship of sampling objects and shows an obvious practical effect as the sampling results has a strong representation and the image data was reduced to 1.758% of the original image volume, which can effectively improve the efficiency of the subsequent image processing.