how can i use RGBDSlam with image

edit

After you brings up RGBDSLAM by using Kinect or other RGBD sensors, you can use rqt to check what are the input topics of RGB and depth image for RGBDSLAM.

Let's say the topic names are /rgb_image and /depth_image, then you need to publish images in your dataset to /rgb_image and /depth_image. You can do it by reading the images from OpenCV and publish it to topics. (You might need this ref.)

Due to some problem I cannot indicate the precise topic name. But I think you can find out by yourself.