August 24, 2006
|
|
"Webcam" Technical Comparison
August 24, 2006 |
Specification Camera Model
|
|
Resolution Active megapixels Sensor chip Pixel size (um) Dynamic range (dB) Sensor SNR (dB) Sensitivity (lux) Color Frame rates (FPS) Sub resolution Image window Data format Shutter (msec) Trigger mode Lens mount IR filter Interface |
HRWC M12 2048x1536 3.1 CMOS 1/2" MT9T001 
3.2x3.2 61 43 ? 24 bit RGB 35 @1024x768 70 @ 640x480 200 @ 320x240 1600x1200 1280x1024 1024x768 800x600 640x480 Yes RGB 0.04 ~ 750 external C-mount standard USB2.0 |
HRWC M1 1280x1024 1.3 CMOS 1/2" Micron MT9M001 5.2x5.2 62 +3 +3a 24 bit RGB 20 @ 1024x768 40 @ 640x480 90 @ 320x240 1024x768 800x600 640x480 Yes RGB +9 option C-mount standard USB2.0 |
DFK 31AF03 1024x768 0.78 CCD 1/3" ICX204AK
4.65x4.65 +2 +3 0.5 24 bit RGB 15 @ 1024x768 n/a Yes? RGB 0.1 - 30K option C-mount option Firewire/1394 |
Logitech 4000 Pro 640x480 0.31 CCD 1/4" ICX098BQ
5.6x5.6 n/a n/a < 1 24 bit RGB 30 @ 640x480 1280x960 (interpolated) No JPEG +9 n/a S-mount removable USB1.1 |
Kodak MDS100 640x480 0.31 CCD 1/3" Kodak n/a n/a n/a 10 24 bit RGB 30 @ 640x480 No TIFF 0.1 - 100 n/a C-mount removable USB1.1 |
|
Notes : 1) A feature (or data) not being available is designated by 'n/a'. 2) All of the cameras in the above table operate in the interline, progressive scan mode. 3) These cameras also incorporate a standard Bayer filter : 
Note that the basic color information is contained in a cell with 4 pixels, each pixel having an associated primary color filter. Thus this information is collected over an area of 2x2 pixels. A pixel is defined by a CCD or CMOS sensor element, with dimensions on the order of microns (see above table). As the eye is most senstive to green light, 2 green filters are included in each color cell. For TIFF and JPEG output formats the "effective" R, G and B 8-bit values - defining a 24-bit color for a specific pixel - are calculated per an algorithm based upon nearest neighbor pixel values. Thus the output color per pixel is an approximation to the actual color, as would be obtained by collecting the R, G and B component information at a single pixel location. It can be expected that an algorithm designed for a consumer point-and-shoot digital camera (or a webcam) to give the best visual effect for a wide range of images, may not necessarily be the best algorithm for processing closely controlled high-resolution images from a microscope. But, the level of sophistication allowed by the latest microprocessors dedicated to these computations makes using a consumer priced digital camera on a microscope a very attractive alternative to a dedicated, scientific quality camera. For the raw RGB output format 8-bit R, G and B data is available for Bayer processing with a user defined algorithm, external to the camera. This allows a great deal of flexibility in matching the 24-bit color camera to a specific application. The Imaging Source, distributor of DFK cameras in the USA, has prepared a White Paper
describing Bayer color filtering along with a summary of algorithms that can be used to calculate the 24-bit pixel data for specific applications.
4) For comparison, this is an image (from a Kodak MDS100 with 25mm lens) of display pixels : 
for a Sony Trinitron monitor. It shows the same RGB information contained in the Bayer cells (4x) shown above.Note the inversion of neighboring RGB triplets for efficient packing of the display pixels and also the increased size of the blue element, to compensate for the eye's reduced response to this color as compared with green and red. 5) An introduction to CMOS imaging, as applied to photomicroscopy, is available on the comprehensive 'Molecular Expressions' website. It includes a diagram of the components forming a complete CMOS sensor : 
The details for the
semiconductor fabrication are shown by the SEM image of a 'pixel cell' (photo diodes and associated control transistors) array forming an imaging sensor.
The transition from CCD to CMOS sensor technology has been documented by Micron, the manufacturer of the CMOS sensors for the HRWC cameras in the above table. Micron also describes the CMOS technology (2006) in some detail, along with the basic components included in a digital camera. 6) The USB2.0 serial interface handles data at up to 480 Mbps, which is 40x faster than the original USB1.1 interface. IEEE1394/Firewire has a similar data rate of 400 Mbps (1394a). The higher data rates allow a viable interface with high resolution (above 0.31 mpixel) imaging sensors and digital cameras. 7) The specifications for the lens mounts (per the above table) are as follows: - C-mount: 25mm diameter with 32 TPI thread and rear flange to sensor distance of 17.53mm - CS-mount: similar to C-mount but with rear flange to sensor distance of 12.5mm Thus a CS-mount lens can be converted to a C-mount by using a spacer of 5mm. - S-mount: M12x0.5 metric standard, 12mm diameter with thread pitch of 0.5mm Although the S-mount allows the use of very compact lenses, the quality of the image obtained is generally inferior to the larger C-mount optics. 8) The relationship between lens format and the size of the image that the lens produces at the surface of an image sensor can be misleading. The lens format (as used in the above table) is related to the diagonal size of the intended image sensor according to: Format 1/4" Diagonal
~ 4 mm1/3"
~ 6 mm1/2"~ 8 mm
For example, a 1/2" format lens is designed to be used with a sensor having a 1/2" specification; but this sensor has a 8.0 mm, not 12.7 mm, diagonal. 9) It should be noted that typical imaging sensors are ~ 5-10x smaller than full frame 35mm film, for which the equivalent diagonal is 43.3 mm. Therefore, when interchanging lenses with these sensors the camera for which the lens was originally designed must be taken into consideration. For example, if a high performance lens designed for a 35mm film camera is adapted to mount on a C-mount "webcam", any sensor smaller than a 35mm film frame captures only the central portion of the image projected by the lens into the 35mm frame area. This results in a cropped field of view (FOV). Conversely, a 35mm camera would require a lens with a longer focal length to achieve the same FOV and thus the term Focal Length Multiplier (FLM) is used to describe the equivalent FOV for an imaging sensor. Hence the FLM for a lens mounted with a 1/2" sensor is 43.3/8 ~ 5.5x, and the cropped FOV is ~ 0.18x the film FOV. It can now be seen that a 200mm telephoto lens designed for a 35mm film camera becomes a 1100mm lens for the 1/2" sensor. In selecting a lens, other than one specfically designed for an imaging sensor per the associated lens formats (see above), some industry standard frame sizes (mm) and diagonals (mm) to be considered are :
Camera 35mm still Frame 36 x 24Diagonal 43.316mm cine10.26 x 7.49
12.78mm cine4.9 x 3.6
6.1Super8 cine7.35 x 5.4
9.2
|
