Kevin George Harding

An intelligent video framework for homeland protection

This paper presents an overview of Intelligent Video work currently under development at the GE Global Research Center and other research institutes. The image formation process is discussed in terms of illumination, methods for automatic camera calibration and lessons learned from machine vision. A variety of approaches for person detection are presented. Crowd segmentation methods enabling the tracking of individuals through dense environments such as retail and mass transit sites are discussed. It is shown how signature generation based on gross appearance can be used to reacquire targets as they leave and enter disjoint fields of view. Camera calibration information is used to further constrain the detection of people and to synthesize a top-view, which fuses all camera views into a composite representation. It is shown how site-wide tracking can be performed in this unified framework. Human faces are an important feature as both a biometric identifier and as a method for determining the focus of attention via head pose estimation. It is shown how automatic pan-tilt- zoom control; active shape/appearance models and super-resolution methods can be used to enhance the face capture and analysis problem. A discussion of additional features that can be used for inferring intent is given. These include body-part motion cues and physiological phenomena such as thermal images of the face.

Absolute measurement using field-shifted moire

Phase shifted moire interferometry is one of the most effective tools for obtaining a full-field depth map. The major draw back of the technique is the two pi ambiguity which limits the measurement depth range to one fringe or requires the counting of fringes across the image. In either case, only a relative measurement is obtained, no information is available about the absolute distance to the camera. By moving the moire projection system (field shifting) the period of the moire pattern is changed allowing extraction of absolute depth information. We have built an instrument, employing field shifted moire to produce a full field depth map with 12 bits of depth resolution. The performance and applications of this instrument are discussed.

Shiny parts measurement using color separation

In manufacturing, inspection and measurement systems have long been desired to be able to measure as many kinds of parts as possible without treating the surface. Specifically, measuring shiny parts has been a big challenge for optical metrology because of double-bounced light -- a phenomenon that light can be reflected from an area to another on the surface. The unwanted light will result in higher noise and can even make the measured results unacceptable. Traditionally, a polarizer is placed in front of both the light sources and the camera. After properly adjusting the polarizer in front of the camera, the double bounced reflected light can be blocked to some degree while the normally reflected light can go through. By this way, the extra reflections can be reduced but not totally eliminated. This paper presents a new method to totally eliminate double bounced light. Here, color light sources are used to illuminate the part and multiple cameras are used to measure different areas. Each camera views through an appropriate color filter so that only a certain color light is seen. The measured results from all cameras are then merged together to create the complete image. This method is more efficient than the traditional solution that uses polarizers. Both measurement principle and some results are given.

Computational approach for optimal sensor setup

Optical metrology has been widely used in quality inspection. However, the sensor setup during part inspection is often done in an ad hoc way. This leads to unnecessarily high signal dynamic range. Consequently, optical sensors do not have sufficient light dynamic range capabilities especially for shiny surface measurement. We present a computational approach for optimal sensor setup that takes into account the sensor/part interaction to decrease signal dynamic range and to increase model coverage for structured light or similar optical inspection systems. First, we transform the signal dynamic range issue into a distance problem in a spherical map. We then present novel algorithms on the spherical map to search a near-optimal sensor orientation. Based on this near-optimal orientation, we use a gradient method to obtain the free-optimal solution that gives the lowest possible dynamic range. Experimental results demonstrated that under the optimal orientation, there is lower signal dynamic range and better model coverage. Future work on extending this method to multiple sensor planning, sensor design, and stage design for large part inspection is also discussed.

OCT for industrial applications

Optical coherence tomography (OCT), as an interferometric method, has been studied as a distance ranger. As a technology capable of producing high-resolution, depth-resolved images of biological tissue, OCT had been widely used for the application of ophthalmology and has been commercialized in the market today. Enlightened by the emerging research interest in biomedical domain, the applications of OCT in industrial inspection were rejuvenated by a few groups to explore its potential for characterizing new materials, imaging or inspecting industrial parts as a service solution[3]. Benefiting from novel photonics components and devices, the industrial application of the older concepts in OCT can be re-visited with respect to the unique performance and availability. Commercial OCT developers such as Michelson Diagnostics (MDL; Orpington, U.K.) and Thorlabs (Newton, NJ) are actively exploring the application of OCT to industrial applications and they have outlined meaningful path toward the metrology application in emerging industry[3]. In this chapter, we will introduce the fundamental concepts of OCT and discuss its current and potential industrial applications.

Mobile, contactless, single-shot, fingerprint capture system

In some applications such as field stations, disaster situations or similar conditions, it is desirable to have a contactless, rugged means to collect fingerprint information. The approach described in this paper enables acceleration of the capture process by eliminating an otherwise system and finger cleanup procedure, minimizes the chance of the spread of disease or contaminations, and uses an innovative optical system able to provide rolled equivalent fingerprint information desirable for reliable 2D matching against existing databases. The approach described captures highresolution fingerprints and 3D information simultaneously using a single camera. Liquid crystal polarization rotators combined with birefringent elements provides the focus shift and a depth from focus algorithm extracts the 3D data. This imaging technique does not involve any moving parts, thus reducing cost and complexity of the system as well as increasing its robustness. Data collection is expected to take less than 100 milliseconds, capturing all four-finger images simultaneously to avoid sequencing errors. This paper describes the various options considered for contactless fingerprint capture, and why the particular approach was ultimately chosen.

Latest optical methods for industrial dimensional metrology

Applying optical methods to a simple inspection problem, such as the presence or absence of an assembly component, is likely the least restrictive of any machine vision application. The setup, alignment, and components can be made very simply by using any feature which appears different if the component of interest is present. Gaging to verify proper construction, however, can be a much more complicated task. To take advantage of the capability of machine vision and other optical technologies, some flexibility in part position, orientation, and speed of presentation may be desired. This paper will review the types of methods available from optical methods for industrial dimensional metrology. The emphasis of the discussion will be 2D and 3D gaging applications.

Surface sensitivity reduction in laser triangulation sensors

Laser triangulation sensors offer a simple, non-contact and fast solution to measure displacement, position, vibration and thickness. However, these sensors are prone to target surface sensitivity since they rely heavily on a uniform back-scatter of the laser spot. Here, sources of measurement noise including surface texture, speckle, beam deflection and asymmetry are discussed. In addition, a few solutions using dithering as well as the beam shaping to reduce surface sensitivity are explored. It is shown that a simple ditherer would induce additional error and a solution is suggested to compensate for it.

High-speed moire contouring methods analysis

Structured light methods, including moire contouring have been demonstrated both technically and commercially as a useful tool for a class of 3D contouring applications. Popular analysis methods such as phase shifting, Fourier analysis, and simple fringe counting all have their strengths and weaknesses. Phase shifting has proven to be a powerful tool for complete surface mapping with good performance, but requires multiple images for the analysis. Fringe counting and Fourier methods offer good speed, but are more sensitive to noise factors. The trends in manufacturing toward tools for in-process measurements has lead to a variety of methods for high sped methods including color moire, multi-camera/sensor systems, and combinations of methods such as stereo with moire methods. This paper provides a comparison of some of these methods with particular emphasis on sources of noise and system error functions. Specific systems discussed include single image fringe counting using gradient methods, multiple camera systems using phase analysis, color encoding methods, motion based systems and image multiplexing methods. The analysis will attempt to separate accountable systematic error sources from random noise and the degree of participation of each in the data analysis methods.

3D imaging using a unique refractive optic design to combine moire and stereo

Phase shift measurement on a Moire image is a very effective approach for gathering a full-field 3D data. The limitations of this approach include: (1) The data is gathered over multiple images as the Moire pattern is phase shifted (one of the gratings is shifted for each of the multiple images). Thus, the data gathering can be effected by things such as motion blurring due to the requirement for multiple images. (2) The phase measurement has a two pi ambiguity which makes it difficult to analyze data with step discontinuities. To eliminate the need to take multiple images as the grating is shifted, we have developed a refractive element system that simultaneously produces multiple Moire images of the same scene. The system is adjusted so that each of the simultaneous images provide a different Moire phase. From these multiple simultaneous images, accurate subfringe information can be extracted using standard phase calculating techniques. An added advantage of this optical design is that the images have a stereo disparity which is a function of the distance from the lensing system. This stereo disparity can be used to eliminate the two pi ambiguity problem that plagues other phase calculating techniques. This presentation reviews the optical arraignment that provides the multiple simultaneous Moire images and presents a mathematical description.