Rather than stripes or geometric patterns, the Kinect uses a projected IR dot pattern and an astigmatic lens (i.e. Even though this method is only capable of capturing the deflection from a single plane in any given 2D view, a 3D mesh of the target’s entire 3D surface can be assembled by rotating the target through 360° and capturing multiple surface maps covering its entire surface.Īs noted previously, the Microsoft Kinect is perhaps the most widely used structured light system.
Beginning with a planar surface as the reference, this method measures the deflection from this reference plane. It is more of a “surface mapping” method, in that it illuminates a surface, and maps the topographical variations of the surface rather than the volume of a body. It is important to understand that Structured Light imaging is not typically used to measure volumetric density (such as the MRI or CT scans used in medical imaging). We are assuming that the camera has not been moved for this hypothetical example. You may notice that only the front (or “upper”) half of the sphere is reconstructed on the right side of Figure 5. The phase shifted patterns are projected sequentially onto the object, and after a phase unwrapping operation, the resulting difference images are compared with a reference plane to extract the surface features of the original object.įigure 5: Phase shifted projections and the resulting 3D construction. Typically, three (or possibly more) images with a known phase shift between the images is used. The Phase Shift method utilizes a set of grayscale images with the pixel intensities defined by a sine wave pattern as shown in Figure 2. For the sake of this discussion, we will look only at the Phase Shift method in more detail. In addition to these methods, there are also hybrid methods that combine more than one method to optimize the results for a particular surface or target.ĭetailed discussions about the wide variety of methods is beyond the scope of this article. Others include continuously varying rainbow patterns, grayscale or color stripe indexing, grid indexing (using pseudo random binary dots or 2D color coded dot arrays, for example), and others. This type of pattern is often used for the Phase Shift sequential projection method - just one of several methods that have been developed. In the example shown in Figure 1, a regular sine wave intensity pattern has been illustrated.
The rounded surface of the ball distorts the stripes, and the distorted image is captured by a camera for analysis and object reconstruction. The direction and size of the pattern distortions are used to reconstruct the surface topography of the target object.įigure 1: A regular striped pattern is projected onto the ball. When the camera views the pattern from one (or more) different perspectives, the surface features of the target distort the pattern, as shown in Figure 1. The concept is quite simple: a known pattern is projected onto a surface. Originally developed for gaming and consumer electronics, the Kinect – and other systems using similar technologies – are finding their way into a wide range of practical applications. One well-known example of a structured light system is the Microsoft Kinect. Structured Light Imaging captures the 3-dimensional topography of a surface using specific, (often modulating) patterns of light, and a 2D imaging camera. One method that is rapidly growing in popularity (and utility) is Structured Light Imaging. Several different 3D methods have been developed and applied to machine vision. Our “stereoscopic” eyes and cognitive processing abilities provide our 3-dimensional awareness, but stereovision isn’t the only way to extract 3-dimensional information. We live in a 3-dimensional world, and many application areas can benefit from 3-dimensional awareness. As the collective needs of our civilization become more sophisticated, the 2-dimensional imaging methods of the past impose limitations on what we can perceive and understand about the world around us.
0 kommentar(er)
