Security Camera Image Sensors: CMOS, Pixel Size, Low Light, and WDR

The image sensor is a central part of a security camera, but its name or megapixel count cannot tell you how the finished camera will perform. The lens, exposure settings, lighting, image processing, compression, and installation all contribute to the recording. Evaluate the sensor as part of that complete imaging system.

For a useful comparison, consider daylight, darkness, moving subjects, strong backlighting, and the distance at which details matter. A camera that produces an attractive still image in a showroom may behave differently when someone walks through a dim entrance or a vehicle passes with headlights on.

What the image sensor does

The lens focuses light onto an array of light-sensitive elements. Those elements convert incoming light into electrical signals. Readout circuitry and analog-to-digital conversion produce data that the camera processes into an image. The camera then encodes a sequence of images for transmission or storage.

The lens forms an optical image on the active sensor area.
Light-sensitive pixels collect charge during an exposure.
The sensor and associated circuitry read and convert the signals.
Image processing handles tasks such as color reconstruction, white balance, and noise reduction.
The encoder creates the video stream used by the recorder or viewing software.

Recorded video can usually be paused to inspect a frame, and supported software can export stills or clips. The detail available in an export comes from what the camera actually captured and encoded. Enlarging a frame does not restore information lost to blur, poor exposure, or compression.

CCD and CMOS

CCD means charge-coupled device; CMOS means complementary metal-oxide-semiconductor. Both are technologies used to make image sensors. They differ in how charge and image information are read and integrated with other circuitry. Neither term by itself determines whether the camera sends analog video or an IP stream.

CCD sensors were widely used in earlier surveillance cameras and other imaging equipment. They should not be described as a technology invented by a particular modern camera brand or as guaranteeing good images in every condition. Their cost, readout arrangements, power requirements, and manufacturing characteristics contributed to changes in the market.

CMOS sensors are common in contemporary security cameras and can integrate useful readout and processing functions efficiently. However, there is no universal claim that every CMOS sensor uses one hundred times less power than every CCD. Compare the actual devices and finished-camera power consumption.

Sensor features such as selective readout or windowing are also different from the camera generating multiple encoded streams. A main stream and substream depend on the complete processing system. Choosing CMOS is a reasonable starting point for many current products, but sensor type alone is not a quality ranking.

Understand manufacturer technology names

Sony's Exmor family includes different sensor architectures. Sony's explanation of Exmor and Exmor R distinguishes the back-illuminated structure used in Exmor R. Do not assume every sensor carrying the Exmor name has exactly the same design.

STARVIS identifies Sony sensor technology developed for security-camera applications, including back-illuminated designs. The generation and specific sensor still matter. A label does not promise that a complete camera can produce an identifiable moving face without adequate light.

Live MOS is a sensor-related term encountered in other imaging products; it is not a universal security-camera quality grade. Likewise, “Starlight” should not be treated here as a generic camera mode supplied by a video-enhancement software company. Dahua, for example, uses the term in its surveillance product literature for low-light imaging capabilities.

Names such as Axis Lightfinder and Hikvision DarkFighter or ColorVu describe manufacturer-specific imaging approaches or product families. They can involve combinations of sensor, lens, exposure, illumination, and processing. They are not necessarily software switches that can be added to any existing camera without hardware changes.

When comparing these names, ask for recorded examples and the exact model's specifications. Use equivalent illumination, target movement, shutter settings, and recording quality. A marketing comparison using different exposure times can conceal important motion-blur differences.

What WDR helps with—and its limits

Wide dynamic range, or WDR, helps a camera represent bright and dark parts of a scene together. A common example is a person inside a doorway with bright daylight outside. Some implementations combine exposures or sensor readouts with different characteristics; others use different processing techniques.

WDR can improve the visibility of shadowed subjects while preserving some bright-area detail. It does not guarantee identical clarity everywhere, remove all glare, or make a dark scene usable without light. Strong reflections, direct sun, and optical flare can remain problematic.

The Axis WDR white paper discusses artifacts such as ghosting when moving objects appear in different positions across combined exposures. Test moving people and vehicles, especially where headlights or changing light levels are involved. The highest advertised dB number is not automatically the best forensic image.

Related features have different jobs. BLC, or backlight compensation, adjusts exposure to favor a subject that would otherwise be too dark. HLC or highlight compensation can reduce the influence of bright regions in supported implementations. Neither should be confused with physically removing a reflection from a lens.

Automatic white balance helps render color under different light sources. Smart IR can adjust infrared illumination or exposure to reduce overexposure of nearby subjects, depending on the product. Region-of-interest encoding can allocate more data to a selected area, while three-dimensional noise reduction uses information across frames and may affect moving detail. Test each feature instead of assuming more processing always improves evidence.

Sensor size and optical-format labels

Labels such as 1/4-inch, 1/3-inch, 1/2.8-inch, or 1/1.8-inch describe historical optical-format classes. They are not literal measurements of the active sensor diagonal in ordinary inches. An inch remains 25.4 mm; the optical-format naming convention does not redefine it as 16 mm.

For precise work, use the manufacturer's active sensor width and height or its published field of view with the supplied lens. A nominal half-inch-type sensor may have an active diagonal around 8 mm, but that is an approximate format relationship, not a universal arithmetic conversion for all products.

All else equal, a larger sensor can provide a larger light-collecting area. With the same focal length, a larger active image area generally gives a wider field of view. But “all else equal” is important: pixel count, pixel architecture, lens aperture, read noise, and processing can change the comparison.

Also check lens compatibility. The lens must cover the sensor's active area and provide adequate optical performance. A larger sensor does not help if the lens cannot form a usable image across it or the final camera crops the area for a particular mode.

Resolution and useful detail

Common video format	Pixel dimensions	Approximate pixel count
720p HD	1280 × 720	0.92 MP
1080p Full HD	1920 × 1080	2.07 MP
UHD 4K / 2160p	3840 × 2160	8.29 MP

These are common formats, not universal recommendations or the maximum possible surveillance resolution. The useful choice depends on field of view and the target. A small display does not make missing recorded detail irrelevant if an investigator later needs to inspect an exported frame.

More pixels can help distinguish smaller features when the lens, exposure, focus, and compression preserve them. They do not automatically produce a proportionally better image. A high-resolution sensor behind a poor lens or with an excessively slow shutter may still deliver unusable moving faces.

Increasing pixel count on a fixed sensor area usually reduces pixel pitch. Smaller pixels may collect fewer photons individually under comparable conditions, but sensor architecture and processing also matter. Noise is not adequately explained by saying that neighboring pixels simply “interfere” with each other; photon statistics, readout, gain, and other factors contribute.

Minimum illumination and lux

Lux measures illuminance, not a complete sensor-quality score. Camera specifications may quote a minimum illumination for a particular output level under stated lens aperture, exposure, gain, and image-processing conditions. A smaller quoted minimum-lux number can indicate greater low-light capability only when those conditions are comparable.

A camera can appear very sensitive by using a long exposure, but moving subjects may blur. Compare footage at a shutter speed appropriate to the task, not merely the lowest number on a datasheet. Read whether the measurement is for color or monochrome and whether supplemental illumination is used.

There is no universal rule to choose “eight to fifteen times” the expected sensitivity, and ranges such as 0.01–0.1 lux do not guarantee a usable nighttime image. Measure or assess the actual scene and test the required subject under its least favorable normal lighting.

Claims of zero-lux operation generally involve active illumination or a different imaging technology. An ordinary visible-light camera cannot obtain scene information from no photons. Higher gain amplifies the signal and noise; it does not create light that never reached the sensor.

Match the camera to the task

Application	What to prioritize	What to verify
Outdoor overview	Coverage, sufficient target detail, environmental suitability	Daylight contrast, moving objects, nighttime lighting
Indoor overview	Field of view, obstructions, useful pixel density	Lights-off conditions and window backlighting
Night recording	Low-light performance at the required exposure	Moving faces or vehicles, IR glare, color needs
Wide-angle coverage	Pixels distributed across the scene	Detail at edges and the farthest relevant target
Access or plate-related view	Task-specific geometry and exposure	Actual recognition performance and supported integration

Fixed sensor-size prescriptions for these categories are unreliable. A warehouse can be darker than an outdoor entrance, and a small interior checkout can need more detail than a broad exterior overview. Access-control or plate applications also need suitable software, angles, and illumination; high resolution alone does not supply recognition.

Storage is governed by encoded bitrate and recording time rather than a fixed file size for 720p or 1080p. A continuous 4 Mb/s stream uses about 1.8 GB per hour in decimal units, regardless of the resolution label. The encoder may need a different bitrate at a different resolution to preserve useful quality.

Night recording and pixel size

Larger pixels can offer advantages in photon collection under comparable conditions. A larger sensor with a sensible pixel count may therefore be worth evaluating for low-light work. But do not discard a higher-resolution model without testing: improved sensor design, lens aperture, and processing may produce a better overall result.

Infrared illumination can substantially improve monochrome nighttime images. Losing color information does not mean losing all spatial detail. If clothing or vehicle color matters, provide suitable visible light or select a camera that can capture reliable color under the site's actual illumination.

IR distance is not a guaranteed identification distance, and a quoted 100-meter illuminator range says little by itself about pixels on a face. Check beam coverage, reflective objects, exposure, motion blur, and the target's position. Nearby walls or dirty covers can create glare that overwhelms the intended scene.

WDR can be helpful in a mixed-brightness night scene, but it may introduce motion artifacts or reduce low-light performance in some configurations. Try supported settings and inspect the saved footage. Automatic enhancement should not be assumed to improve every scene.

Decide whether the task requires detection, recognition, or identification. Clothing, height, and build can provide context, but they do not guarantee that a person will be uniquely identified. Do not intentionally sacrifice required facial detail on the assumption that an intruder will be easy to recognize anyway.

Weather and installation

Sun angle, headlights, seasonal foliage, rain, and reflected light affect an outdoor image. Sensor dynamic range, optics, exposure, processing, and placement all remain relevant. BLC or HLC cannot compensate for every poor mounting position or optical flare.

Environmental specifications are separate from sensor type. Check ingress protection, operating and startup temperatures, impact resistance, cable sealing, and any heater requirements. A camera may need to tolerate severe cold or heat at one site but not another; do not apply −40°C or +50°C as a universal outdoor requirement.

Direct sun can heat an enclosure above ambient temperature. Ventilation, sunshades where approved, mounting orientation, and the manufacturer's instructions matter. Plan cleaning and inspection because water spots, dust, and scratched covers can undermine an otherwise capable sensor.

Manufacturer names and final selection

Image sensors are supplied by specialist semiconductor manufacturers, and camera makers may use different sensors across models or production revisions. A camera brand is not proof that its sensor was manufactured in-house. Use documented specifications rather than an assumed supply-chain relationship.

The source listed consumer examples such as Xiaomi C200, IMOU Cube, and TP-Link Tapo C225, alongside Dahua and HiWatch models for other settings. Those names are examples from the original discussion, not a verified current ranking or a recommendation for U.S. availability. Exact variants, support, integration, and imaging performance require separate checking before purchase.

Compare candidate security cameras with the same target, distance, lighting, motion, and recording settings. Ask whether the resulting footage meets the task, whether the recorder preserves it, and whether it remains useful after export. Sensor size, pixel count, and low-light claims are valuable inputs, but the recorded result is the final test.