Get the quality image to indentify face with CCTV camera.
Can 4K IP camera capture facial detail for identification? How far my camera can see? People may want to understand this question when they choose a Security camera.
First of all, we need to consider pixel density rather than pixel. Pixel density is measured by pixel per meter(PPM) or pixel per inch(PPI). The parameter we find in a camera's specifications such as 1920*1080 is pixel, not pixel density. It needs to divide into camera's field of view to get the density.
You may ask why we only consider single Horizontal pixel, but not whole picture 1920*1080. The result is same if we also include the height of view in the format. we only measure single line to make the it simple.
The higher pixel density, the more facial detail you have. It is similar as flood control with the sandbags on the dam. The more sandbag on the dam, the density is higher, and the dam will become stronger.
John Johnson of U.S. Army Night Vision lab has developed the pixel density requirement for surveillance applications. It needs around 196 PPM for recognition, 130ppm for classification, at least 65PPM for detection. To idetify the face, you may need higher density such as 260PPM even 330PPM
The below is the automatic calculator to measure how far a camera can see with a specific pixel density. The FOV will be converted between the width of camera's view and you will have the result about how far you camera can see basing on your goal of your camera setup. Some cameras have motorized zoom function which allows the camera to see farther with narrow coverage area, and increase pixel density as well.
Let's make a example, 4K (8MP) camera, 60 degree FOV, with 260ppm to achieve better recognition. The result is 14 meters. It means your target area should not be 14 meters far from the camera installation point if you want better recognition. You can simple increase the density to 350ppm to have identification. The distance reduces to 11 meters. If 11 meters is overkill, you can pick up lower resolution camera, such as 4/5 megapixel ip camera or increase the horizontal to 80 degree. One trick thing is 12 megapixel camera only double up the pixel density than 2 megapixel camera, not 6 times. Why? because part of pixel increasing has been deployed to vertical direction, making the camera see wider instead of clear. Multi sensor IP camera can simple increase the pixel density. It adds the sensor to horizontal area. Pixel density is metric elements to evaluate what you really can get from your camera such identification or detection.
The second element we should consider is the shooting angle. If you try to turn down a camera to see a shorter distance in order to raises pixel density, You may properly capture people's head instead of the facial detail. The optimal installation angle should be less than 30 degree from ceiling. Of course, the more directing shooting to the face, it has better chance to capture facial detail, such as eye level camera setup. Other elements including illumination, slow shutter, frame per second. With insufficient light, the person has best vision can turn into blind. This is the same as camera. The slow shutter is trick, most of the IP cameras are enable the slow shutter to accumulate light. The consequence is you get ghost image when the object moving in low light condition. You can simple disable the slow shutter. Yes, we are not making movies, why needs 25 frame per second. It is because only the key frame can deliver the facial detail, the more frame speed can reserve more possible key frame.
GOLBONG have release several megapixel IP camera for facial purpose, including height strip camera which you can install on the door frame. Eye level 180 panoramic camera, and even smoke detection covert camera. All those cameras can generate sufficient pixel density for facial identification, and those cameras can be install to eye level position to capture the face even when people wear a hat. The TF card memory storage allow you to back up the video on the camera or you can choose the TF card as primary recording, retrieve the video by network connection. The motion recording can trigger the notification and push a message to smart phone when people pass thru the camera. It saves a plenty of time to locate the person of interest. The below you can find the products detail.