In this paper, the hybrid active contour model driven by optimized local pre-fitting image (OLPFI) energy was proposed for fast image segmentation. Since the proposed model combines the local pre-fitted image (LPFI) function with region-based characteristic and the optimized edge indicator function with edge-based characteristic, the proposed model can not only correctly segment images with intensity inhomogeneity or weak boundaries, but also can deal with images whose target has large noise interference inside. The two pre-fitting functions are computed before curve evolution, so the computational cost of the proposed model reduces greatly. Furthermore, the evolution curve can adaptively select the motion direction according to image information because of energy sign function. Regularized function used in the proposed model can accelerate the segmentation speed of the proposed model. Moreover, the addition of the regularization term and length term helps to improve segmentation efficiency of the proposed model and eliminate the effect of sensitive parameters in traditional penalty and length terms. Experimental results show that the proposed model has great efficiency in segmenting images and great robustness on initialization and parameters.

在本文中，提出了一种由优化的局部预拟合图像（OLPFI）能量驱动的混合主动轮廓模型，用于快速图像分割。由于该模型结合了具有基于区域特征的局部预拟合图像（LPFI）函数和具有基于边缘特征的优化边缘指示函数，该模型不仅可以正确分割强度不均匀或弱边界的图像，而且可以处理目标内部噪声干扰较大的图像。这两个预拟合函数是在曲线演化之前计算的，因此所提出模型的计算成本大大降低。此外，由于能量符号函数，演化曲线可以根据图像信息自适应地选择运动方向。所提出模型中使用的正则化函数可以加快所提出模型的分割速度。此外，正则化项和长度项的加入有助于提高所提出模型的分割效率，消除传统惩罚项和长度项中敏感参数的影响。实验结果表明，所提出的模型在分割图像方面具有很高的效率，并且在初始化和参数方面具有很强的鲁棒性。