An official website of the United States government.
<OL> <LI> Three-Dimensional Plant Health Characterization<BR> Three-dimensional Vision and in general Multi-View Machine Vision techniques had been applied in industry: checking for can labeling, tire markings, etc. and in Agriculture, they had been used for automatic tractor navigation in the field, automatic obstacle avoidance, "coarse" row-mapping of plant height. So far the applications rely on monochrome or color features. Most objects studied were "closed or convex-shaped" as contrast to "open architecture with radiation-absorbing multi-elements" found in plant canopies. Thus there is a need to develop an instrument providing more plant canopy details for phenotype screenings of plants (used for biomass production, for example poplar and switchgrass), to complement existing X-ray systems for the below-ground part. In addition to the 3-D information, there is also the need to acquire the "spectral" components to estimate plant growth & health. <LI>Three-Dimensional Motion Recovery of Nanomotors<BR> Recent research goals are to use nanomotors as self-propelled chemicals delivery vehicles into specific targets inside humans, animals and plants for disease control and possibly cure, therefore their motion mechanics must be ascertained more precisely so that their optimal design or trajectory control can be eventually performed. This research objective is similar to Objective 1 as this work is also based on Multi-View Machine Vision of an object, except that the object was static in Objective 1 and therefore we were dealing multiple static images, while in Objective 2 the object is always in motion, so we will be dealing with image sequences of the same object and here we have severe optical configuration restraints as we are eventually dealing with a microscope system. Here the goals are to recover the 3-D dimensions of such structures and their kinematic parameters such as linear and rotational velocities, with a future research goal of eventually doing a reverse analysis of this resulting dynamic motion back to its possible model of the dynamic forces acting on the nanomotor being analyzed so that its optimal design or trajectory control can be achieved.
NON-TECHNICAL SUMMARY: Situation-Problem-Purpose With the fields of biomass and plant health research expanding, there is a need for a research tool to be developed for use in the "phenotype-screening" of plants at the canopy level, requiring 3-dimensional coordinates information of leaves and stems as well as their spectrometric characteristics in order to quantify their growth and health status. A triple-camera dual-wavelength spectral imaging system will be developed to scan and reconstruct the 3-dimensional progress of plant stresses within the plant canopy, concentrating at first on biomass and chlorophyll quantification on selected plants such as cotton, poplar and switchgrass. Hardware and software technologies developed for and from this Stereo Spectral Imaging system will also be applied to establish guidelines on when and how a 3-D motion recovery can be successfully achieved on macro-structures and by extension to nano-structures using video recordings from a single camera.
<P>
APPROACH: Three-Dimensional Plant Health Characterization. Using a triple camera system XB3 from PointGrey Co. and BAE custom designed dual-wavelength filter system, a Stereo Spectral Imaging (SSI) system will be constructed for chlorophyll and biomass quantification of selected plants (cotton, poplar and switchgrass). This XB3 system has 3 cameras positioned on a line and PointGrey already provides software for synchronizing multiple cameras and 3-D data recovery. BAE has to develop filter switching and spectral calibration procedures and "stitching" software to create a mosaic of the whole canopy. A proof-of-concept experiment will be performed whereas selected leaves of a plant canopy are treated with herbicide. Such plant is then scanned with the SSI system to show 3-D and time-wise progress of herbicide stresses. This SSI system will be used in conjunction with spectro-radiometers to verify existing research models for light absorption through plant canopies,as this SSI system can quantify geometrical orientations of given leaves with respect to incident radiations. Three-Dimensional Motion Recovery of Nanomotors. At present, there is no reliable procedure to precisely measure actual dimensions of individual nanostructures, thus in this project we still need to stay in the macro world so that accurate dimensions of studied objects can be measured and served as "reference" data to compare with the dimensions "recovered" from the Multi-View Machine Vision system. We'll start with 3-D recordings of macro objects in motion (translational and rotational motions combined) using 2 video cameras set up on opposing sides of the "moving" plane of the object (please note that the final application is for a microscope system where the object's motion is restricted to a plane with limited motion in the depth dimension). The proposed approach is to start with 2 cameras enabling reconstruction of 3-D objects, next we remove the information from one camera, thus we have to algorithmically replace that lack of information by "trying" different possible values of the 3-D objects like lengths of nanomotor arms or angle of rotation, etc.,let's say by a Genetic Algorithm (GA) tool for example. So by starting from data from one actual 2-D video frame, we can use a GA tool to predict what the next 2-D frame should be like from the 3-D architecture that the GA tool "assumes" at that time, and check the "predicted" and "actual" frames to see if all pixels, especially edge pixels "fall" within the right X-Y coordinates system using a "least-squares" performance measure function to drive the GA tool towards convergence. We also will try different pixel resolutions because in actual nanomotor recordings the pixel resolution is usually poor. In conclusion we want to establish some guidelines on when and how a 3-D motion recovery can be successfully done on macro structures and by extension to nano structures.