Arthritis is viewed as a disease caused as a result of inflamed joints. Arthritis is the most familiar element of disability in the world. Now a days in Africa there are six common Arthritis. Those are Rheumatoid Arthritis, Osteoarthritis, Gout Arthritis, Juvenile Arthritis, Ankylosing, Spondylitis, Psoriatic Arthritis. Though X-ray scans alone are insufficient to detect the type of arthritis easily, Image processing can improve the diagnosis. Therefore, there is a need for effective model development to identify knee arthritis. The previous study neglects identify both Osteoarthritis, Rheumatoid Arthritis and Gout Arthritis together. In this thesis we develop image processing models for identification of arthritis types like Rheumatoid Arthritis, Osteoarthritis and Gout Arthritis. Because those are frequently occurred arthritis types. In this study, we have done image preprocessing(image size normalization 224x224,image denoising and RGB to Gray scale conversion), image segmentation(canny edge detection), image feature extraction(Convolutional neural network), classification of arthritis(Convolutional neural network) techniques were conducted. Therefore, we develop feature extraction and classification model using CNN. Finally, we used Softmax to classify images as normal, Rheumatoid Arthritis, Osteoarthritis and Gout Arthritis. We examined the performance of CNN model for both feature extraction and classification techniques. While we use RGB color72.5% was registered at testing time .The model Achieved a diagnosis accuracy of 78% for training and 74% for testing to detect and classify the Arthritis images while we use Gray Scale. Better performance was registered while considering the gray scale images of knee Arthritis as an input to the CNN model.