For the requirements in practical application of indoor service robot, a real-time object recognition system is proposed which integrates 3D point cloud segmentation and local feature matching. First, it does fast and effective object detection based on 3D point cloud. Then, the areas of objects in color image are located using the result of object detection, and objects are identified using the approach based on SURF feature matching. The experiments show that the system could well meet the real-time and reliability requirements of indoor service robot's object detection and recognition.