Stand E01, Vision UK@Photonex, The Ricoh Arena, Coventry, UK, 15 & 16 October, 2014

STEMMER IMAGING will be showing an imaging system capable of sorting and classifying robotic fish and also demonstrating high resolution inspection of pixels on an LCD screen at Photonex.

Fish classification is carried out using the Manto recognition and classification tool from STEMMER IMAGING’s Common Vision Blox machine vision toolkit, with images produced by a JAI Elite EL-2800C-GE2 2.8 Mpixel GigE Vision camera with advanced Sony CCD technology (the ICX674 2/3″ progressive scan imager) for excellent sensitivity and reduced smear. The system can differentiate between similar sized robotic fish swimming around at random.

Manto is a vision tool designed for applications where other methods of pattern recognition fail. It uses a neural technology based on the principles of support vector machines. A critical component of this is a new type of multi-scale pre-processing filter which allows the software to adapt to a wide range of different tasks automatically. It is different from other pattern recognition approaches because it allows, among other things, the recognition of organic forms and textures that show a high level of distortion or that are hard to reproduce. The system learns to identify the patterns of interest from a set of training images and then calculates a confidence factor for its classification choice for each test image. Unlike traditional neural networks, Manto, can use any number of training images, bringing an ability to classify objects with an accuracy not previously possible.

The high resolution demonstration will feature a 9 Mpixel GigE camera to inspect a 7” LCD screen. The system inspects for hot and dead pixels and clearly shows the red, green and blue sub pixels on the display.

The camera used here is the Manta G917 colour camera from Allied Vision Technologies, equipped with a Schneider Unifoc lens for high optical performance with very low distortion. The Manta G-917B/C includes an 1” Sony ICX814 CCD sensor with EXview HAD II technology offering excellent image quality.  Measurements are carried out using the Sherlock machine vision platform from Teledyne DALSA. Sherlock’s is a versatile measurement environment, with a rich suite of proven vision tools and advanced algorithms that can be applied to a wide variety of automated inspection applications. Standard tools within Sherlock include: search (pattern finding), ID readers, character readers (OCR), measurement and counting tools, contour tools, texture and colour classifiers. A variety of speciality tools have also been included to simplify difficult inspection tasks.