Face in face recognition procedure is mainly influenced by what factors?
2022-10-30 00:00:00
face image starts to affect the recognition effect, face image will affect the recognition rate.Non-professional face-recognition cameras common rules of minimum recognize faces for more than 60 * 60 or 100 * 100 pixels.Within the prescribed image size, the algorithm is easier to improve accuracy and recall rate.Image size is reflected in the actual application scenario is face the distance from the camera.
2: image resolution
as it is difficult to identify low image resolution.Comprehensive image resolution image size, directly affect the camera distance.Now 4 k camera to see faces is the longest distance between 10 meters, 7 k camera is 20 meters.
3: light environment
too light or too dark light environment will affect the effect of face recognition.Can from the camera to bring the function of fill light or filter balance light effects, can also use optimization algorithm model light images.
4: fuzzy degree
the actual scene mainly focus on solving the motion blur, motion will often cause the face relative to the camera motion blur.Some cameras have the function of anti fuzzy, and under the limited cost, optimization model considering algorithm to this problem.
5: shade degree
facial features, facial without sunscreen, edge image is the best.In the actual scene, many face will be hat, glasses, masks and other obstructions, this part of the data need to decide whether to retain the training according to the algorithm requirement.
6: collect Angle
face relative to the camera Angle was right in the face of the best.But the actual scenario it is often difficult to capture is face.Because this algorithm model needs to training includes left and right side face, upper and lower side face data.Industrial camera placed on the construction point of view, face and the camera Angle to be fulfilled within the scope of the algorithm to identify the requirements.