No abstract

The display of three-dimensional angiograms can benefit from the knowledge of quantitative shape features such as tangent and curvature of the centerline of vessels. These can be obtained from a curve-like skeleton representation. If connectivity and topology are preserved, and if geometrical constraints such as smoothness and centeredness are satisfied, it is possible to estimate length, orientation, curvature, and torsion. It is also required that no part of the original object be left unrepresented. An efficient method for the identification of such shape components is developed. First, a suitable representation is obtained using a voxel coding approach to yield connected and labeled unit-thick paths. The desired features are estimated from a smoothed version of the skeleton produced by a moving average filter. The computational cost is linear, of the order of N object , the total number of object voxels contained in the binary volumetric data. The method is also shown to be robust to boundary noise. Examples are discussed.

This paper introduces a two-phase algorithm to extract a center-adjusted, one-voxel-thick line representation of cerebral vascular trees from volume angiograms coded in gray-scale intensity. The first stage extracts and arranges the vessel system in the form of a directed graph whose nodes correspond to the cross sections of the vessels and whose node connectivity encodes their adjacency. The manual input reduces to the selection of two thresholds and the designation of a single initial point. In a second stage, each node is replaced by a centered voxel. The locations of the extracted centerlines are insensitive to noise and to the thresholds used. The overall computational cost is linear, of the order of the size of the input image. An example is provided which demonstrates the result of the algorithm applied to actual data. While being developed to reconstruct a line representation of a vessel network, the proposed algorithm can also be used to estimate quantitative features in any 2-D and/or 3-D intensity images. This technique is sufficiently fast to process large 3-D images at interactive rates using commodity computers.

Surface defects of cathode filaments of microwave magnetron would cause magnetron failure and scrapped microwave systems. Therefore, surface defects on cathode filaments must be carefully inspected. Conventionally, filaments are manually and visually inspected using their amplified images under an optical microscope. This is because automatic defect detection of cathode filaments is a challenging problem. The difficulty comings from its complex surface shape with multiple turns of high curvature spiral circles, which occlude each other. Such complex shape prevents capturing of sharp focusing images, which are essential for a computerized automatic detection algorithm. Further, the variable nature of production defects complicated the process of automatic defect detection task. To solve these problems, this paper proposes an automatic defect detection method to deal with issues related to complex shapes containing occlusions as well as high curvatures, particularly for the quality inspection of spiral shaped cathode filaments. The method includes a novel digital scanner, which sequentially brings all sections of the filament sides into sharp focusing of the optical imaging system. The method also employs multiple optical systems to imaging multi-sides of the spiral filament. The computational algorithm primarily uses line-detectors. In an evaluation experiment, the proposed method was used to automatically inspect over 14 million cathode filaments. Experimental results indicate that its false negative rate was 0.0065%, and its false positive rate was 6.83%. This indicates that the proposed method could successfully detect all kinds of surface defects at over 99.99% accuracy. It reduces the workload for manual inspection from 100% down to 93.17%, over an order of magnitude reduction. Further, the efficiency of the proposed method is 70 spiral filaments per minute, satisfying the requirements of online quality detection of existing manufacturing lines of filament cathodes.

Aiming at solving the problems of low-efficiency and high-cost of existing auto-focusing methods for microscopic imaging, an on-line auto-focusing method is proposed and developed using dual photodetectors. Firstly, two identical photodetectors are placed in symmetrical positions with equal distance away from the focal plane in the image space. One is in front, the other is behind the focal plane. The photodetectors convert the light intensity signals originating from the sample into two electric signals, and their difference is proportional to the amount of defocus. Using a precalibrated curve between the differential voltage signal and the amount of defocus, the actual distance of the current sample from the focal plane can be measured in real time. The sign of the differential signal determines the direction of defocus, i.e., before or behind the focal plane. The sign and magnitude of the differential signal is used to control the direction and number of steps of a motor, which drives the carriage to move toward the focal plane and bring the sample into sharp focus, realizing the on-line automatic focusing of the microscope. The results of experiments show that the system can achieve efficient and stable automatic focusing. Under 40X objective lens, the effective working range of the proposed method is ±40um.Within this working range, the time required to achieve auto-focusing is 0.90702s, with the majority time, i.e., 0.907s spent on mechanical movement while 0.02ms for sensing the defocus. The major advantage of the proposed method is its high efficiency defocusing sensing at an updating frequency of 50k Hz. Further, it is easy to implement and with a low cost. Therefore, it is reasonable to expect that the proposed autofocusing method would have wide applications include classroom teaching, laboratory microscopic imaging, and inline product quality inspection.