Imaging technology

We can mine valuable information from images

Computer Vision and image analytics is used to acquire vision sensor data and process it for making decisions. Vision sensors refer to a broad range of sensors that can deliver data as images and include cameras, lasers, lidar, thermal sensors, x-rays and others. For some applications, human screening operators currently review live image data, for example in the case of CCTV or luggage screening with x-ray, however without automation it is impossible to quickly interpret and act on vision sensor data that can generate millions of images in a few hours. The automation of vision sensor data analytics is not a trivial task. The complexity of software increases exponentially as we move from application environments with fixed visual geometry, e.g. industrial machine vision, to unconstrained environments such as outdoor areas where the visual appearance of an object depends on multiple factors such as lighting levels, varying angles of view, overlap with other objects, scene geometry variation, and others. The problem becomes further challenging if real-time analytics is needed. The biggest challenge is to minimise false positive and false negative detections and ensure that analytics software can make detailed measurements for object detection and recognition. Our expertise includes, but is not limited to:

• Optical image analytics using areas scan and line scan cameras, for applications where either the object or the sensor is moving at speed, e.g. railway track inspections using vision sensors mounted on high speed trains, drone based surveying and condition monitoring of land based assets
• Optical image analytics using cameras inserted in an enclosed area with no fixed geometry (e.g. medical laproscopy)
• Optical image analytics using cameras inserted in an enclosed area with fixed geometry (e.g. robots used with cameras to inspect sewers, water pipelines etc.)
• Thermal image analytics for applications where high temperature may suggest a potential hazard, e.g. hot train wheel after bearing failure on a moving train or electrical power lines
• Laser profile measurements for applications where the 3D profile of the object must be acquired and compared with a template to detect differences, e.g. rail wear measurements by comparing laser profiles of new and old rails, or quality control of manufactured objects
• Lidar point cloud measurements for applications where a three dimensional map of the environmental surveyed must be created using Lidar point cloud data acquired with time of flight measurement technology, e.g. 3D profiling of roads, railway tracks, tunnels, structural assets
• X-ray image analytics of monochrome images as in medical applications or as dual energy colour coded images in the case of security scanners for detecting threats in luggage, cargo and parcels
• CT and MRI image analytics for medical applications
• Radar imaging for distance measurement or object recognition or for collision avoidance, e.g. autonomous driving applications, water level measurement in canals, robotics