Image guidance plays a major role in precision radiotherapy. Because conventional X-ray image guided techniques used in radiotherapy deliver additional doses to patient, an image guided radiation therapy (IGRT) technique using conventional video cameras that does not cause harm to the patient was investigated in this study. Three conventional video cameras were installed in different directions to acquire real-time images. The backgrounds of the images were deleted using an adaptive background removal algorithm designed in this study, and then the region of interest was obtained. Real-time patient positioning information could be achieved using an oriented FAST and rotated BRIEF algorithm that we modified. Respiratory signals were derived from the real-time images of the conventional video cameras using an algorithm we designed. A plastic patient model immobilized with a thermoplastic shell on a simulation couch was used to verify the accuracy and feasibility of our method. Test results show that the position and angle offset errors identified with our system are less than 1 mm and 1°, respectively. A respiratory motion simulator is designed and used to test the respiratory signal acquirement algorithm we designed. The time and amplitude conformity of the respiratory signals between the setting and acquirement is better than 96%. In our laboratory tests, the developed IGRT technique is shown to satisfy the requirements of image guided radiotherapy.