A saliency detection model is proposed based on Log-Gabor filtering and saliency map fusion optimization of stereoscopic images, in which the image saliency is detected by the planar image saliency combined with the texture and depth features. First, the left view saliency map is calculated by the improved graph-based manifold ranking algorithm. Second, the left view texture features and the depth features from stereoscopic images are extracted, and the texture and depth saliency maps are computed by the Log-Gabor filtering method, respectively. Third, the above three saliency maps are integrated into a stereoscopic (3D) saliency map by the weighted linear combination (WLC) method. Finally, the 3D saliency map is enhanced by the center-bias factor and visual acuity. The experimental results on a public eye tracking dataset show that the proposed model possesses a good detection performance and is superior to the existing 3D visual saliency detection models.