Process of finding human knee in image based on multiple weighted thresholding and histograms of gradients

Osteoarthritis is the most common joint disease, affecting about 12% of the population. It mainly affects the weight-bearing hip and knee joints but is also commonly found in other locations (spine, shoulder, elbow, and fingers). In Osteoarthritis, the articular cartilage is gradually destroyed, and its mechanical properties are altered, typically manifested externally by pain, frequent inflammation, and swelling.As part of the clinical needs of physicians, regular monitoring and recording of knee conditions are advisable during the follow-up and treatment of the disease. This paper proposes a process to obtain an image of the knee using a camera mounted on a sliding stand. To ensure the repeatability of the image acquisition, a software process is presented that allows finding the knee in the image. The found knee can be used in the unattended device to move the camera to an optimal position corresponding to the centre of the patient’s knee.