Testing IP Thermal Camera (Axis Q1910)
By Benros Emata, Published Jun 01, 2010, 12:00am EDTIn the last 6 months, interest in IP based thermal cameras have increased with Axis, FLIR and Pelco announcing relatively low cost product offerings.
In this test report, we investigate how well the Axis Q1910 thermal camera works in the field, integrating it with Exacq's VMS and with Agent Vi's video analytics. We also contrast the performance of the thermal to Axis's Q1755 HD camera to better understand tradeoffs.
We explore the following questions:
- What is the setup/configuration required to use the camera?
- How far can the camera see?
- What is the maximum FoV width to detect humans?
- How does the camera's performance compare to a megapixel camera?
- What palette or color profiles should be used to optimize viewing?
- How does the camera impact the use of video analytics?
- What limitations should be factored in for using the camera?
Below is a 1 minute preview of the premium report inside:
Physical Overview & Web Interface
The following screencast covers the physical form factor and web interface of the Axis Q1910.
Key points include:
- Similar in form factor to the Axis Q1755
- Q1910 has a fixed 13mm lens
- Image sensor has a resolution of 160x128
- In the web interface, image can be scaled up to 720x576
- Web interface has a palette option that changes the thermal color scheme on the camera image
Thermal Palette Comparison
In the following screencast, we discuss and preview the variety of palette options the Q1910 interface provides. This test was conducted during nighttime at 0.3 lux, in an outdoor environment. We compared each thermal color scheme to see what would most compliment an operator's view in detecting a human.
Day Versus Night Color vs Thermal Operation
In this screencast, we contrast the performance of the Q1910 thermal camera with the Axis Q1755 day/night camera in daytime and nighttime scenarios. The objective is to find the tradeoffs and options for each use within the given scenario.
Key points include:
- During the day, the subject's heat profile was similar to vegetation in the environment, which reduced the impact of the thermal but did not affect the color camera
- In daytime, the color camera had an advantage to make out details than the thermal camera
- At night (0.3 lux), the thermal camera had an advantage to human identification than the color camera
- Although thermal camera could detect human activity, details could not be determined from image
Maximum Distance Test
We tested how the resolution of the Q1910, combined with the thermal performance, impacted the ability to detect humans at different distances. We conducted this test at night with a light level of 0.3 lux, in an outdoor environment. We had our subject in camera view stop at pre-defined points every 100 ft, up to 500 ft.
Key points include:
- At 200 ft, 63.8 ft wide field of view, subject's head started to blend with background (body temperature was similar in color to background environment temperature)
- At 400 ft, 128.6 ft wide field of view, subject not as bright, getting difficult to detect
- At 500 ft, 161.3 ft wide field of view, subject was small and with colors blending with background, made it a challenge to detect
- Using the Black-hot palette displayed contrast better than the default White-hot
Video Quality Samples
You may download sample video clips (90 MB) from the Q1910 thermal camera that we have been referencing throughout this report.
Using With Analytics
In this screencast, we examine the performance of the Q1910 thermal camera's use with video analytics.
Key points include:
- For night performance, thermal camera does a better job at seeing a subject than a color camera under natural light
- For analytics, thermal cameras require less pixels per foot, with the additional potential benefit of better nighttime performance
- Drastic changes in light and shadows often create issues with analytics
- Thermal camera image does not change under direct, shining light, nor with moving shadows
- Different temperatures between a tree in foreground and parking lot in background increases the tree's visibility which may cause false positive results in analytics
- Depending on the environmental conditions, human subject may have similar body temperature to ground, causing possible missed analytic detection
