The new model can reserve the object information effectually, eliminate the noise, augment image gradient, and pose stronger constringency effect.

• 新模型能有效地保留目标信息，消除噪音，增大图像梯度值，具有更强的收敛效果。