Human-computer interaction based on hand skeleton and associative rules

The problem of identifying the needs of patients who cannot speak or wear gloves that transmit neural signals to indicate the need is an important field, especially since computer vision fields provide many options for identifying gestures, which are considered a means of communication. Understanding and responding to human signs requires hand gesture interpretation, which is a fundamental HCI ability. This is because human-machine interaction must feel natural. This lets computers record and analyze human gestures. A system that accurately and reliably recognizes hand gestures is desperately needed since humans use them to communicate. Computer vision has several uses where gloves or sensors aren’t needed. After detecting and tracking the hand with the Landmark Point, joint characteristics were retrieved. Based on the retrieved features, examined the data and generated key hand movement classification criteria. After extracting the features from the previous step, the rules are organized using association rules that give good results by reducing many rules with low confidence and support. The proposed method for recognizing numbers and a number of other gestures, depending on the built rules gave an accuracy of 100% for the close distances between the hand and the camera up to 180 cm and for the far distances of 200 cm it has an accuracy of 99.8%. The method also showed excellent effectiveness for recognizing the gestures used in an uncontrolled environment, as well as the correct recognition of the gesture in real-time, at a rate of 10 Ms.