Software for 3D Stereo Vision

scorpionxii-64x64

Scorpion Vision XII

3D Stereo Vision is the extraction of 3D information from digital images, such as

obtained by a Scorpion 3D Stinger Camera. By comparing information about a scene from two vantage points, 3D information can be extracted by examination of the relative positions of objects in the two panels.

The Scorpion Vision Software tool # 97 – Locate3D calculates the 3D position x,y,z of two corresponding points from two 3D Calibrated Images. The accuracy using 3D Stereo Vision is down to 0.001 mm with Scorpion Vision Software.

Image Rectification – Image rectification is supported by tool component STC-0084 StereoRectify3D. This transforms two 3D Calibrated images into two stereo rectified images where all epipolar lines are parallel to the horizontal axis and the corresponding points have identical vertical coordinates.

2D Image Calibration is performed by the Scorpion tool # 15- Calibrator.

calibrator2d

The Scorpion Calibrator tool ensures sub-pixel accuracy down to 1/20th of a pixel. The foundation for the most accurate Scorpion 3D Stinger for Robot Vision systems.

resolution

System Resolution is 0.07 mm with a pixelresolution of 0.8 mm

Active Stereo Vision signified that energy in the form of light is projected into the scene.

The Scorpion 3D Stinger Camera is an industrial stereo vision camera in an IP-64 enclosure. The camera is supporting passive and active stereo vision. The camera is supplied with POE or 24 volt and an ethernet connection. Active stereo vision is supported by the means of integrated white or IR high power leds, structured illumination with different pattern based on red or IR lasers.

is-2016-0022-scorpion3dstingerforstubmeasurement_small

Scorpion 2D and 3D Stinger Cameras in Aluminium Industry

The Scorpion Stinger SMARTcard strobes the laser and the LED illumination. This is important to meet the Laser Safety requirements for machine vision system. Strobing also improves the ambient light protection and the Signal to Noise ratio of the projected pattern. The SMARTcard also serve as trigger interface to encoder on a conevyor.

Structured Illumination – the Scorpion 3D Stinger Camera supports multiple projection patterns to solve the following challenges:  create features on featureless surfaces, solve the basic feature correspondence problem. The following patterns are supported: Single Line, Multiple Lines and RPP – Random Projection Pattern.

Standard High Quality GigE Cameras from Sony and Basler is used as the main image sensors in the Scorpion 3D Stinger Camera. Resolution from VGA to 5MP is supported with monochrome or color CMOS or CCD sensors optimized for price, speed or image quality.

30 mm and 200 mm Baseline – the Scorpion 3D Stinger Camera is manufactured in two different baseline models. The 200 mm baseline version is designed for Robot Vision and Gauging applications. The typical 3D FOV is a Euro Pallet – W = 800 x H = 1200 mm x D = 500 mm. The 2.3 MPixel model has a 2D x,y resolution better than 0.1 mm and a typical depth resolution 0.3 mm with a good 3D Calibration optimized by One Push Calibration.

One Push 3D Calibration is a concept for automatic fast, accurate and reproducable multiple 3D camera calibration using one single image and a Scorpion Stinger 3D Calibration object.

OnePushCalibration

The automated One Push Calibration provides an optical 3D resolution of 0.04 mm – 1/20th of 0.8 mm which is the 2D pixelresolution. This provide a depth resolution with a 200 mm baseline camera at 1455 mm distance of  Zres = 0.04 * 1455 / 200 = 0.29 mm – the 2D resolution is wellk below 0.1 mm

3D Point Cloud – A 3D image or 3D point cloud is an ordered or unordered sequence of 3 points.

3dpointcloud

Dense 3D Point Cloud generated by Scorpion 3D Stinger for Robot Vision – typically resolution 1mm in all three dimensions x,y,z

Sparse 3D Image Creation – Scorpion Vision can accumulate single, multiple 3D feature points into a 3D Point cloud. The 3D images are often used to locate an object plane with tool 121 – PlaneFit3D.

Single Line Projection is normally used for Laser Triangulation. Scorpion 3D Stinger Cameras for standard Laser Triangulation are available. The Scorpion tool # 166 – LaserProfiler3D provides an easy means to create point clouds from a sequence of images with a single line laser projection.

Multiple Line Projection – a laser with a multiple line projector is used to project structured light on an object or on a plane. The projected lines can be located in 3D with the advanced tool component STC-0064-LocatePaths3D. The sparse 3D from multiple 3D points generators is moved into a 3D Image for further processing.

RPP – Random Pattern Projection – a laser with a random pattern projector is used to project random texture onto an images. The disparity of the projected pattern is ideal for BlockMatching.

rppblended

Intrared RPP Image captured by Scorpion 3D Stinger Camera – the blend images combines RPP texture with the natural texture – ideal solution for objects moving on a conveyor

Dense 3D Image Creation – Scorpion Vision Software supports different means of Dense 3D Image Creation. 3D scanning using 3D Laser Triangulation with a Scorpion 3D Stinger Camera and a Single Line Laser is a basic and proven method when the object or the camera is moving. The tool component STC-0083 DisparityMap3D produces a dense 3D image within a 3D FOV from two stereo rectified images. The method can be used with passive and active stereo vision. A RPP – Random Pattern Projection laser will guarantee features on a featureless surface and improve the applicability of block matching. The STC supports multiple Block Matcher to handle different block matching scenarios.

DenseStereoVision

Our favourite concept image – IR LED Image1 – IR RPP Image 1 – blends to an accurate 3D Model – The rectified HeightMap is ideal for 3D pattern location with TemplateFinder3 with 4DOF

Scorpion 3D Stinger for Robot Vision 2.0 is a product developed and distributed by Tordivel AS.

%d bloggers like this: