The resolution and lens of the camera define the image thatís best for your application. Do you want to view people in a small store or do you want to look at the cars in the parking lot? Do you want to identify a personís face or just count the number of people in an area? Once you know what you want to see you can specify the resolution of the camera and the type of lens you will need. Resolution defines how clear a picture looks. The lens defines how much you will see at one time (the Field of View). This article describes how to select the right camera resolution and lens. Resolution
The higher the resolution the more detail we can see. The amount of resolution required depends on the application. For example, suppose we are looking at a room thatís 40í by 40í square, and we want to recognize a personís face from across the room. In this case we will require more resolution then if we just want to count how many people are in the room. In this example, we may select a camera with 700 x 480 resolution instead of one with 350 ax 240 resolution.
There are two ways to define resolution of a video system. Historically TV was measured in Lines of Resolution; today digital cameras are defined by the number of Pixels in the camera sensor. Both of these measurements are related. TV lines of resolution
are classically defined as the maximum number of black and white lines that can be seen on a monitor. Historically, resolution was measured using a test pattern. It provided a series of converging lines with resolution numbers next to the lines. The idea is to look at where the lines converge. When you get to the point where you are just able to see the black-white pattern converge into all black, you have reached the point of maximum resolution. We still use this old TV standard to define how clear a picture appears on a monitor. If you really want to go through the details, take a look at our previous article Understanding Resolution and Brightness
. Digital Camera Resolution
is determined by the number of sensor elements available. It is defined by the horizontal (H) and vertical (V) number of pixels (i.e. 700 x 480) or by the total number of pixels in the sensor (H x V = 336K Pixels). Equating TV lines to pixels
It would seem that all you really need is one pixel for the black line and another pixel for the white line to equate pixel resolution to TV lines of resolution. But, there are other factors that affect the actual resolution like the Kell factor which says that only 70% of the pixels are actually visible at one time. Most importantly, resolution is in the eye of the beholder, so it depends on how clearly the image gets to the monitor you view it on. How much resolution do we need?
The U.S. Department of Justice has provided some guidelines to school systems so they could select the right resolution camera:
ďFor observation of a camera scene to determine only if a human is in the scene (or to be able to distinguish between a person and an animal), a minimum criteria of 6 horizontal TV lines across a 1-foot-wide object within the scene is used. (In terms of active picture elements, this means that a 1-foot-wide object would cover 8 horizontal active picture elements for each row of picture elements for the height of the object on the camera imager.) For identification of a person by facial features, 16 horizontal lines (21 pixels) of resolution subtending a 1-foot-wide object are needed.Ē
This is almost right. 16 horizontal TV lines (vertical resolution) equates to 21 TV lines of horizontal resolution (using 3:4 aspect ratio). TV line resolution and pixel resolution are related but different measurements. As I mentioned above, you require more than twice the number of pixels to provide the equivalent TV line resolution. More importantly, we need to consider what you will actually see on the monitor. The Real World View
The best way to determine what resolution you need is to try it out with a video camera and lens system. IQinvision did some testing and published their results on their web page. The following test was done by IQinvision.
Methodology: Using an IQeye3 and a 6.5 mm lens they increased distance and took an exposure until the face was unrecognizable. Then the head and shoulders were cropped from the large image. They used Photoshop to copy the face and enlarged it to 92 by 110 (the native resolution of the largest face) using bicubic interpolation. No sharpening or other manipulation was used.
As you can see in the pictures above 42 to 50 pixels per foot is probably the minimum number of pixels necessary to identify a personís face. By the way, if we used the recommendation from the Department of Justice, we come up with a similar number. The calculations are: 2 x 21 = 42/.7 = 60 pixels. How wide an area do you want to look at with the camera?
The field of view determines how wide an area you will see. Do you want to view people in a room or in a parking lot? Do you want to see a persons face or just know that they are there? If you need to identify a personís face, then using the recommendations above, we need a minimum of 50 pixels/ft to identify a personís face.
To determine the field of view, we first select the resolution of the camera and then the lens.
Suppose we have a camera with a resolution of 2048 x 1536, then we divide the horizontal pixels available by the minimum number of pixels/ft to get the total horizontal field of view. In this example 2048/50 = 40 ft. So 40 ft is the maximum field of view required to allow us to identify a personís face. If we just want to see how many people are in the view, the field of view can be over twice as large. Lens Selection
The lens determines what you will see in the video display. You can select a wide angle lens to view objects that are close to the camera, or select a telephoto lens to see objects further away.
If we use a high resolution camera you should select a high resolution or Mega-pixel lens that will maintain the high resolution. We can use an automatic zoom lens or use digital zoom to look at a portion of a wider scene. A mechanical zoom lens can provide 25X or more of zoom. Digital zoom capability depends on the resolution of the camera. For example a camera with 2048 x 1536 of resolution can be zoomed about 16X before the image becomes too pixilated to make out the picture. Selecting the Lens
Select the lens based on how far away the 40 ft field of view is from the camera. The closer to the camera the wider angle (or smaller mm) lens we require, the further away the narrower the angle and the larger the mm of the lens. In our example, a 4mm lens will allow us to view a personís face up to 25 feet away from the camera. A 10mm lens will move the field of view 95 feet from the camera. A lens calculator
can help you calculate the distance to the object, width of the field of the view and lens. You can also get an excellent preview of what an actual lens will look like at the IQinvision web siteís example
of Multi-Megapixel resolution Conclusion
The resolution of the camera and the lens determine what we will see. A higher resolution camera provides better detail. The lens determines the field of view. Both factors work together to provide the view we need for the application.
You can get more information about the other important camera specifications such as light sensitivity and frame rate in the article Understanding Camera Specifications
in our previous newsletter.
Need some help selecting the right lens and resolution, just contact us. We will be happy to help. 1-800-431-1658 (in the USA) or 914-347-2530 (outside the USA) or send us a message