Looking on the bright side
21 March 2017
A whitepaper looks into the causes of image brightness variation in machine visions applications and offers some possible solutions.
Machine vision systems are used to make many critical measurements as part of a quality control approach where it is essential that the measurements are both accurate and repeatable.
Since machine vision measurements are made on the image of the object on the sensor rather than the object itself, it is important to optimise every element of the imaging process to achieve the best possible image.
Illumination is vital in the generation of an image with good enough quality for processing and measurement and LED illumination is the most commonly used in machine vision applications. While the type, orientation and wavelength of the illumination can all play a crucial role in delivering good quality images, stability and repeatability are absolutely essential for a vision system to perform consistently, so undefined variations in illumination are unacceptable.
The Image brightness is a function of the amount of light arriving at the image sensor. Clearly the illumination source plays a critical role in this, but other components in the optical system, and external factors can have an influence on this.
Any imaging system which has thresholding, or is looking for subtle features such as surface defects or colour inspection will benefit from improving the repeatability of the light levels. Consistent illumination allows threshold parameters to be set closer to the background level, allowing the detection of finer features, and increasing the contrast between target features and the background. Applications involving dimensional checking or the detection of gross defects, such as the presence of features, will be more tolerant to variations in illumination intensity.
LED lighting control methods
Continuous lighting: Continuous operation is the simplest mode of LED lighting control. The light is on all the time and the light intensity is proportional to the current supplied to the LED by the lighting controller. The maximum light intensity achievable is 100% of the LED manufacturer’s rating.
Pulsed lighting: Pulsed lighting control offers a number of benefits. In pulsed mode the lighting is switched on only when needed and the controller receives a trigger signal when a pulse is required.
Pulsing makes it possible to freeze the image of moving objects, making it suited to high speed imaging applications. An additional benefit of pulsed operation is that it is possible to obtain more than 100% brightness from an LED by driving it with more current for short pulses.
Intelligent lighting: The intelligent lighting approach, taken by Gardasoft’s Triniti intelligent lighting platform, takes lighting control to a new level by the networking of LED lighting, camera and imaging software to provide an integrated application with a single graphical interface for set up and control.
Issues affecting output
Controlling the light output from the LED is a fundamental requirement for maintaining constant image brightness and there are a number of factors which affect light output. These include:
Age of the light: The light output from an LED will deteriorate over time and, typically, an LED would be replaced when output falls to below 70% of its initial rating. The rate of decrease of light intensity will therefore be a function of how the LED is driven for a given application and this is more difficult to predict.
Temperature of the light: Gardasoft has identified that as LEDs heat up from 25°C to 90°C brightness can drop by up to 40%. This is a huge variation that may not be seen during system commissioning, but which can cause variability during normal running.
Variations in the drive to the light: LED light output is proportional to the current through the device, not the voltage, so all LED device manufacturers specify that current control is advised for efficient use.
Making use of the characteristic profile of output vs current (provided by the light manufacturer) can enable very linear brightness control, even for high levels of overdriving. Accurate regulation of the current is essential, since 1% variation in current can lead to 1% variation in light intensity.
Environmental effects: It is possible for the environment of the machine vision system to affect LED output. Another factor to consider is condensation when the LED is cold. This may affect light output due more to temperature (cooling) effects than absorption of the light by the condensed liquid.
While accurate regulation of light intensity is critical in maintaining image brightness, there are a number of other factors which affect the amount of light reaching the image sensor. These include:
Ambient light: Most machine vision systems try to exclude ambient light from the camera view to remove this as a variable factor. However, this is not always possible. Reducing the exposure time will reduce the effect of ambient light. Machine vision lighting generally needs to be proportionally brighter, possibly by using the overdriving feature of a lighting controller. Keeping the exposure time the same but increasing the lighting brightness allows the iris to be closed, also reducing the effect of ambient light. Sometimes it is possible to separate the two types of light. For example, if visible light is needed for the normal working environment, it may be possible for a robot vision to use near infra-red light. Using narrow range visible lighting, such as LEDs or filtered conventional lighting prevents the visible light affecting the infra-red imaging.
Variations in lighting and camera exposure can also affect imaging brightness. When using pulsed lighting, it is essential to get the timing of the lighting pulse and the camera exposure perfectly aligned in order to optimise image brightness. If they are not in alignment, then the image will appear dark or in the worst case, no image will be seen. By adjusting the lighting pulse delay, the two can be brought into alignment. The Triniti intelligent lighting system, for example, enables the user to set up the timing for a whole machine vision system, with cameras and strobe-mode lighting, all from one place.
As well as dust, dirt, liquids or vapours sticking to the LED, it is also possible for them to adhere to the surfaces of the lens system used, reducing light throughput, which will, in turn, will have a detrimental effect on overall image brightness. Compensation can be achieved either by increasing the camera gain or in the software processing of the image.
This can include adjusting thresholds and other parameters based on the operator’s judgement. However, starting with an image which is not at the ideal brightness means that the image processing could result in less dynamic range in the final image. It is usually better to maintain the brightness of the original camera image, rather than compensate in the image processing with possibly compromised performance.
When considering variability in image brightness it is essential to think about the tolerance in brightness that is acceptable – either within and individual system or from system to system. Having decided what measures to take to achieve this tolerance, careful consideration should be given to what extra performance or reduced maintenance times could be achieved by taking additional measures to maintain constant image brightness.
Contact Details and Archive...
Most Viewed Articles...