Tuberculosis (TB) is a disease caused by a bacterium called Mycobacterium tuberculosis. Basically there are two types of TB, latent and pulmonary TB. It symptoms include persistent cough for more than two weeks, cough with blood, night sweat, tiredness, fever, chest pain, lymph node enlargement. It treatment depends on severity level such as mild, moderate, severe and very severe and this can be obtained from TB diagnosis such as chest x-ray image, nucleic acid amplification test and culture and sensitivity. The main aim of this work is to determine severity level of TB using image processing techniques. Chest x-ray (CXR) images of TB patients were obtained from Google image. Firstly, the CXR images were enhanced from Red Green Blue (RGB) to gray image (GI) form. Secondly, GI were decomposed, convolved, compressed and filtered these processing is called image degradation. Thirdly, These GI were restored and binarized. Threshold value ≥ 53 was obtained to classify the severity level of the infected image Furthermore, average error was calculated and system performance obtained. This work revealed that image 1, 2, 6 and 11 has a total pixel value 11067, 6735, 9256 and 3894 respectively ≥ 53 which indicates that the % of the infected areas as 99.62%, 99.52%, 96.70% and 82.78% respectively which shows the severity level are very severe. While the image 3, 4, 5, 8 and 9 has a percentage of infected areas 73.13%, 72.90%, 76.80%, and 76.84% respectively which indicate the severity level as severe. Image 7 has a percentage of infected area 51.21 which shows the severity level as moderate. This research work revealed that image processing is a suitable technique for determine the severity level of TB.