The traditional visual attention mechanism is complex and rough-detection for visual saliency detection indoor red-green-blue (RGB) image. In order to overcome these defects, a new fast visual saliency object detection method based on fusion depth information on indoor RGB image is proposed. A certain scale image is obtained by sub-sampling and pyramid-quantization to reduce the spatial resolution of the images so as to decrease the computational complexity. The intensity, red-green and yellow-blue three-channel features visual attention mechanism significant detection model is proposed to acquire saliency map. The saliency growing strategy is proposed to acquire the precise saliency region in the saliency analysis. The fusion depth information is utilized to detect the objects in salient region. The feasibility and effectiveness of the algorithm is verified in indoor detection experiments.