Introduction
The demand and use of IP cameras, for video surveillance requirements, is growing rapidly; and the time is right to discuss how important the megapixel rating of a camera is, for good quality video capture. This article is split into two parts for easier readability.
The popular view, in the digital photography world, is that the more the megapixels (supported by the camera) the better is the quality of the captured image or video. This simplistic view ignores the subtleties behind image capture, and this note sheds light on two important components of a camera’s content capture mechanism: the sensor and the lens.
Sensor
The sensor is the most neglected component of a camera’s content capture mechanism, when evaluating the specifications of an IP surveillance camera (or digital compact camera). There are three parameters that need to be checked when comparing sensors:
Technology
Sensors can either be CCD or CMOS, and each technology has its relative advantages: high light sensitivity and rich colour representation, in the case of CCD; and low power consumption, low cost, and fast read-out in the case of CMOS. The two technologies have evolved significantly in the last few years; so much so that there is little to differentiate between the two, today.The WDR (Wide Dynamic Range) or HDR (High Dynamic Range) CMOS sensor is a development of interest to the video surveillance industry. The ability of WDR/HDR CMOS technology to handle extreme variations of bright and dark lighting conditions in the same visual frame, makes it ideal for security applications.
Size
This detail usually shows up as ⅓”, or ½”, or ⅔”, in the specifications of the IP surveillance camera. At the outset, it is safe to say that the larger the figure the better will be the quality of the captured image; provided the sensors are of the same generation and of similar quality. The importance of the size of the sensor is that larger sensor sizes have larger photosites (photosensitive diodes that act as photon capturing devices); which results in the capture of more light, with less noise. The outcome is images that are clearer, sharper, and brighter. The reasons IP surveillance camera vendors go in for smaller sensor sizes are primarily cost and form-factor.There is another implication, of the sensor size, on image quality; and that is on pixel size. We shall explore this facet when we discuss resolution. The following link provides more information on camera sensors, for those interested:http://www.cambridgeincolour.com/tutorials/camera-sensors.htm
Resolution
Marketing departments in camera companies have almost exclusively focused on “megapixels” as a determinant of image quality, for cameras; much like a leading processor company, a few years back, focused on “gigahertz” as a determinant of performance, for CPUs. The megapixel norm is attractive, because it is a number and easily understood by consumers; but, just like the gigahertz norm, it is now coming up short against the expectations of consumers.In compact digital photography, 6 megapixels is widely quoted as the resolution beyond which viewers fail to register a significant improvement in image quality. However, this number is a gross simplification; because the actual determinant of image quality is pixel size. Provided the sensor is sufficiently large, one can keep increasing resolution and see a corresponding increase in image quality (finer detailing). The threshold for good image quality seems to be a pixel size of > 3 µm2.
How do we arrive at pixel size? Simple: [Area of the sensor in µm2 / No. of pixels]. Let us take an example of an IP surveillance camera featuring a ⅓” sensor and ~2 Megapixel (MP) resolution:
Each sensor size corresponds to a specific dimension, and the common sensor sizes and dimensions are given below:
| Sensor Size | Dimension (mm) | ||
| Diagonal | Width | Height | |
| 1/3” | 6.00 | 4.80 | 3.60 |
| 1/2” | 8.00 | 6.40 | 4.80 |
| 2/3” | 11.00 | 8.80 | 6.60 |
| 1” | 16.00 | 12.80 | 9.60 |
The dimension of the ⅓” sensor is 4.8 mm x 3.6 mm, which translates to 4,800 µm x 3,600 µm. The area of the sensor is then calculated as 17,280,000 µm2.
Dividing 17,280,000 µm2 with 1,920,000 (1600 x 1200 resolution sensor) gives us a figure of 9.0 µm2 as the size of the pixel; which on the face of it looks to be a very good figure.
However, if we consider an often missed specification in IP surveillance cameras – digital zoom support – we quickly realize that anything more than 3X digital zoom will result in a poor quality image.
We will look at how much resolution is required for surveillance situations in another article.
