Developing accurate machine vision is one of the final obstacles that the wood processing industry now faces. In this research, radiant heating of lumber followed by scanning for knots with an infrared image analysis system (thermography) was performed. Wood specimens containing two different knot diameters were heated with 1500 watts of radiant heat for 60, 120, and 180 seconds. Temperature measurements were taken from both the knot and the surrounding clear wood at 60, 120, 180, and 240 seconds after the specimen was removed from the radiant heat. Knots were readily detected by the infrared scanning system. As the heating periods increased, the required cooling time to detect a significant temperature difference between knots and clear wood decreased. The shortest processing time at which a significant temperature difference was detected between knots and clear wood was 240 seconds, which consisted of heating for 60 seconds and cooling for 180 seconds or heating for 180 seconds and cooling for 60 seconds. This research was done on southern yellow pine. Theoretically, however, any species might be used. This technology is relatively inexpensive, safe, and can be used to accurately differentiate clear wood from knot wood. Furthermore, this machine vision system requires no human interface. If production-line speeds are achieved, this system could be incorporated into grading stations and optimized cutting machine centers (trimmers, crosscut saws, ripsaws, etc.).