1. Research background and significancefatigue testing has great significance and practical value in real life, and it is also a subject worthy of further improvement.
2. Research objectivesThe goal of this design is to utilize MATLAB powerful image processing ability and practical and convenient programming method, through processing contains the human face video image, the recognition analysis facial characteristic, thus obtains the comparatively accurate fatigue condition.
3. System Software Design FrameworkSkin color-based face detection and extraction algorithm flow, the color range extracted in the HSV color space, the skin color range is set to white, the non-skin range is black, and internal fill, can be obtained in the face range of black and white template; human eye location and statistics based on Hough transform; using Prewitt operator for edge detection, The edge curve of the human eye is obtained, and the edge curve of the human eye is obtained from the edge image. The edge curve of the human eye is not very complete, so do the swelling treatment, get the human eye range, convenient to statistics size information; The ratio of the number of frames to the total number of eyes closed in each image is obtained. Judge fatigue.
4. SummaryThis design is simple, but the operation speed is relatively slow, mainly is the Hough transformation computation ratio is large. In order to reduce the computational volume, we can further reduce the human eye search scope. For the face, can be determined that the human eye is generally located in the middle of the head of the position, plus the lower neck and other areas, you can think of the human eye in the upper part of the head, so you can reduce the search range by half, the operation speed can also be accelerated by one.
Design and implementation of a fatigue detection system based on facial feature recognition
