Tuesday’s marvels of engineering: High-Resolution Variable Density Scanning

Industrial machine vision setup requires customization due to a high variety of products, materials etc. Therefore, fast and robust implementation of machine vision applications is critical for short time-to-market.

Integrators of industrial automation lines need machine vision solutions allowing simple integration by an automation engineer (not a machine vision expert), eliminating a requirement for special programming skills or custom algorithm development, and at the same time improving precision and increase robustness of the solution.

The image data collected using conventional approaches includes too much irrelevant data, requiring heavy processing to extract the important information, since a large amount of collected irrelevant data contaminates the relevant information. This results in complicated development by machine vision experts and many development iterations to achieve robust performance.

Focus on What Really Matters

Saccade Vision, an Israel based start-up venture, has developed an innovative three-dimensional machine vision solution for industrial dimensional inspection and vision guided robotics applications. The Saccade Vision solution includes two main components: a unique smart hybrid 3D sensor and flexible and autonomous software. Employing proprietary, patent-pending technology, Saccade simulates the way human vision works, by ‘zooming’ into important information on the measured part and by disregarding the non-useful information.

This unique approach allows the Saccade sensor to perform highly precise and fast three-dimensional measurements and inspection by focusing only on critical locations and increases machine vision robustness on various materials and in different lighting conditions.

The Saccade machine vision solution provides coordinate measuring machine (CMM) levels of precision, resolution, and robustness at a rate of hundreds of measurements per second, making it ideal for the inline process control. The solution can be installed and programmed by a non machine vision expert using an automatic conversion of a digital model of the inspected part (CAD drawing) directly into the machine vision solution.

Based on digital mode analysis, Saccade Vision engine generates an optimal 3D acquisition method, light pattern(s), scanning density for each fine measured feature individually. The engine also ignores locations on the measured samples that are not among the features of interest. This autonomous approach improves the quality of acquired data and aids the cutting of application development times from weeks to just hours.

Measurement in the ‘Blink of an Eye’

The innovative sensor mimics human vision mechanism called saccade (sa-käd). Human eyesight has two main functions: central and peripheral vision. Central vision is associated with focusing on particular objects of interest, while peripheral vision is used when something to the side attracts our attention. Central vision provides us the highest level of details with-in a very narrow field of view, while the peripheral vision has wide field of view and low resolution.

Having the full field of view with the resolution of central vision would require an equivalent of 600 megapixel true color camera. Something that even the strongest computer such as our brain can’t possibly deal with.

Instead our brain analyses the low-resolution image, identifies objects of interest and quickly moves eyes to capture interesting parts with high resolution while ignoring non-important parts of the scene. These quick eyes movements are called “sa-käd”.

By imitating the sa-käd mechanism of the human perception, Saccade is offering machine vision with an unmatched precision and ultimate simplicity. “We believe that to simplify machine vision solution, the quality of information acquired by imaging sensors must be improved and the irrelevant data that ‘contaminates’ the important information shall be eliminated” comments Alex Shulman Co-Founder and CEO.