目前以維度縮減的方法為主,例如主成份分析(PCA),切片逆迴歸法(SIR) 相關的方法等,做影像的分割問題。影像特徵的選取可以在空間域,頻率域及Gabor 的空間-頻率域。 

jImageSeg is a Java-based tool for the supervised and unsupervised image segmentation. The segmentation approaches currently include the principle components analysis (PCA), sliced inverse regression (SIR) and the SIR-related methods. More segmentation techniques and the basic image processing procedures will be added in the future. The feature extraction can be performed in three domains: the space blocks, the Fourier transformation and the Gabor filter banks. The segmented image is displayed by superimposing the resulting boundaries on the input image. [Suspended Work]

 

                    圖例中為胸部超音波影像, 以PCA 做出來的分割結果。

 

Add comment


Security code
Refresh