Keys to Choosing the Best Image Sensor

Keys to Choosing the Best Image Sensor

Image sensors are the key component of any camera and vision system.  This blog summarizes the key concepts of a tech brief addressing concepts essential to sensor performance relative to imaging applications. For a comprehensive analysis of the parameters, you may read the full tech brief.

Download Tech Brief - Choosing the Best Image Sensor

While there are many aspects to consider, here we outline 6 key parameters:

  1. Physical parameters

    Resolution: The amount of information per frame (image) is the product of horizontal pixel count x by vertical pixel count y.  While consumer cameras boast of resolution like car manufacturers tout horsepower, in machine vision one just needs enough resolution to solve the problem – but not more.  Too much resolution leads to more sensor than you need, more bandwidth than you need, and more cost than you need.  Takeaway: Match sensor resolution to optical resolution relative to the object(s) you must image.

    Aspect ratio: Whether 1:1, 3:2, or some other ratio, the optimal arrangement should correspond to the layout of your target’s field of view, so as not to buy more resolution than is needed for your application.

    Frame rate: If your target is moving quickly, you’ll need enough images per second to “freeze” the motion and to keep up with the physical space you are imaging.  But as with resolution, one needs just enough speed to solve the problem, and no more, or you would over specify for a faster computer, cabling, etc.

    Optical format: One could write a thesis on this topic, but the key takeaway is to match the lens’ projection of focused light onto the sensor’s array of pixels, to cover the sensor (and make use of its resolution).  Sensor sizes and lens sizes often have legacy names left over from TV standards now decades old, so we’ll skip the details in this blog but invite the reader to read the linked tech brief or speak with a sales engineer, to insure the best fit.

  2. Quantum Efficiency and Dynamic Range:

    Quantum Efficiency (QE): Sensors vary in their efficiency at converting photons to electrons, by sensor quality and at varying wavelengths of light, so some sensors are better for certain applications than others.

    Typical QE response curve

    Dynamic Range (DR): Factors such as Full Well Capacity and Read Noise determine DR, which is the ratio of maximum signal to the minimum.  The greater the DR, the better the sensor can capture the range of bright to dark gradations from the application scene.

  3. Optical parameters

    While some seemingly-color applications can in fact be solved more easily and cost-effectively with monochrome, in either case each silicon-based pixel converts light (photons) into charge (electrons).  Each pixel well has a maximum volume of charge it can handle before saturating.  After each exposure, the degree of charge in a given pixel correlates to the amount of light that impinged on that pixel.

  4. Rolling vs. Global shutter

    Most current sensors support global shutter, where all pixel rows are exposed at once, eliminating motion-induced blur.  But the on-sensor electronics to achieve global shutter have certain costs associated, so for some applications it can still make sense to use rolling shutter sensors.

  5. Pixel Size

    Just as a wide-mouth bucket will catch more raindrops than a coffee cup, a larger physical pixel will admit more photons than a small one.  Generally speaking, large pixels are preferred.  But that requires the expense of more silicon to support the resolution for a desired x by y array.  Sensor manufacturers work to optimize this tradeoff with each new generation of sensors.

  6. Output modes

    While each sensor typically has a “standard” intended output, at full resolution, many sensors offer additional switchable outputs modes like Region of Interest (ROI), binning, or decimation.  Such modes typically read out a defined subset of the pixels, at a higher frame rate, which can allow the same sensor and camera to serve two or more purposes.  Example of binning would be a microscopy application whereby a binned image at high speed would be used to locate a target blob in a large field, then switch to full-resolution for a high-quality detail image.

For a more in depth review of these concepts, including helpful images and diagrams, please download the tech brief.

Download tech brief - Choosing the Best Image Sensor

1st Vision’s sales engineers have an average of 20 years experience to assist in your camera selection.  Representing the largest portfolio of industry leading brands in imaging components, we can help you design the optimal vision solution for your application.

What are global shutters and rolling shutters in machine vision cameras? How can we use lower cost rolling shutter cameras?

machine vision cameras shuttersWe often are asked the question, “What is the difference between a global and rolling shutter image sensor in machine vision cameras? ”  Although they both take nice pictures, they are very different image sensors with pro’s and con’s of each.  In the end, rolling shutter image sensors cost less, but are not always recommended for moving objects.

In this blog post, we will explain the differences between global and rolling shutter sensors used in machine vision cameras.  Additionally, we highlight how to use a rolling shutter camera capable of  “Global Reset”  providing low cost solutions for some applications with moving objects.

First, let’s explain the differences between rolling shutter vs global shutter image sensors in machine vision cameras.

Global Shutter:  Image sensors with a global shutter allow all of the pixels to accumulate a charge with the exposure starting and ending at the same time.  At the end of the exposure time the charge is read out simultaneously.  In turn, the image has no motion blur on moving objects.  This is given the exposure is short enough to stop pixel blur which is a topic for another blog.
Global shutter image

Rolling shutter:  Image sensors with a rolling shutter do NOT expose all the pixels at the same time.  Alternatively, they expose the pixels by row with each row having a different start and end time frame.  The top row of the pixel array is the first to expose, reading out the pixel data followed by the 2nd, 3rd & 4th row and so on.  Each of the rows start and end point have a delay as the sensor is fully read out.  The result of this on moving objects is a skewed image
Rolling shutter image

What are the Pro’s and Con’s of each type of shutter?

Global Shutter:  
Pro:  Freeze Frame images with no blur on moving objects.

Con:  Global shutter sensors require more complicated circuit architecture, thus limiting the pixel density for a given physical size.  In turn, sensors with a global shutter will have a larger image format driving up lens cost.  The complicated circuits also drive up the overall camera cost and will be more expensive vs a rolling shutter sensor.

Rolling Shutter:
Pro:  Rolling shutter sensors have a simpler design with smaller pixels, allowing higher resolution in a smaller image format allowing use of lower cost lenses. The simpler pixel design results in lower camera costs!.. For example, Dalsa’s 18MP Nano for < $600!

Con:  Image distortion occurs due to the row by row integration and offset on moving objects.  Smaller pixels may also require a higher quality lens which is commonly gauged by the lens Modular Transfer Function (MTF).  This is really dependent upon your application and can be discussed with a sales engineer In turn, there maybe a small trade off to consider.

Is there a way to use a lower cost rolling shutter camera on moving objects?  Absolutely using a Global Reset mode found in various image sensors.

Using a rolling shutter capable of a “Global Reset” such as the AR1820HS found in the 18MP Teledyne Dalsa Nano C4900 camera will eliminate the image distortion.

A typical rolling shutter image sensor as described above exposes sensor rows separately with a delay as depicted below.
rolling shutter mode

Using a rolling shutter with global reset mode, all rows start integrating at the same time as shown below eliminating the image distortion.  It is highly recommended however to use a dedicated strobe and sync with the start of image acquisition.  A gradient in the image brightness from top to bottom maybe seen if not with some pixel blur due to longer row exposure
rolling shutter with global reset mode

A great camera to consider is the 18MP Teledyne Dalsa Nano C4900 camera.  This camera features the ON-SEMI AR1820HS sensor with this capability.  With a price point of < $600, this makes it one of the lowest cost cameras per pixel on the market.

contact us

1st Vision has over 100 years of combined experience and can help you with camera, lens and other peripheral recommendations.  If you have questions regarding the various sensor shutters, please do not hesitate to contact us!

Be sure to read our related blog posts:

What is a lens optical format? Can I use any machine vision camera with any format? NOT!

Demystifying Lens performance specifications – MTF