Corner Pixels Research Articles

Advanced feature point transformation of corner points for mobile object recognition

In this paper, a new method is proposed for recognizing rotated objects based on extracted feature points from the Harris corner detector using object corner points at real time in a mobile environment. With this proposed method, corner points can be rapidly extracted from the input image and object recognition can be conducted by comparing existing feature points. Color values for corner pixels in the feature points of rotated objects are generally varied in accordance with the degree of object rotation. The color values of the feature points are affected by the degree of rotated objects; therefore, there is a possibility that these values can be mixed with nearby pixel values. By analyzing the color values of the rotated pixels, we can extract feature points of rotated objects and use them for object detection to be effectively applied for rotated object detection. Although overall recognition rate is somewhat degraded, using the proposed object recognition with corner information of feature points makes it possible to execute in real time in mobile environments by reducing the amount of computation. Accordingly, this method can be applied to recognizing rotated objects, even in a mobile environment with limited computing capabilities. Experimental results show that the proposed method can provide approximately 96 % accuracy and high performance compared to other methods. Therefore, the proposed method can be adapted to recognize any rotated object with performance and accuracy.

An Efficient Trademark Image Retrieval using Combination of Shape Descriptor and Salience Features

Trademark carries the prestigious values for a particular company so it is very important to distinguish it from the similar context trademark. In this research paper, we propose an efficient Trademark Image Retrieval (TIR) model which is a branch of Content Based Image Retrieval (CBIR). In the proposed system we extract the edge point of a particular image and after this the edge point are evaluated to find the corner pixel from it. For the performance evaluation of the system we use the most commonly used method namely precision-recall. From the experimental result we conclude that the TIR based on shape feature perform better and gives satisfactory result.

Automatic Georeferencing of Images Acquired by UAV’s

This paper implements and evaluates experimentally a procedure for automatically georeferencing images acquired by unmanned aerial vehicles (UAV’s) in the sense that ground control points (GCP) are not necessary. Since the camera model is necessary for georeferencing, this paper also proposes a completely automatic procedure for collecting corner pixels in the model plane image to solve the camera calibration problem, i.e., to estimate the camera and the lens distortion parameters. The performance of the complete georeferencing system is evaluated with real flight data obtained by a typical UAV.

Age Classification based on Corner Pixel Grey Level Co-Occurrences Matrix (CP-GLCM) of TN-LBP

The present paper proposes a novel scheme based on third order neighbourhood LBP (TN-LBP). The present paper observed and noted that the TN-LBP forms two types of corner pixels i.e. top corner and bottom corner pixels. The present paper derived Grey Level Co-occurrence Matrix (GLCM) based on LBP values of Top Corner Pixels (TCP) of TN-LBP and Bottom Corner Pixels (BCP) of TN-LBP. On this GLCM features are derived. Based on these features human age is classified in to child (0 to 12 years) young adult (13 to 30 years), middle age (31 to 50 years) and senior age (above 60 years). General Terms Classification, Image Processing et. al.

Monte Carlo investigation of charge‐transport effects on energy resolution and detection efficiency of pixelated CZT detectors for SPECT/PET applications

Semiconductor detectors are increasingly considered as alternatives to scintillation crystals for nuclear imaging applications such as positron emission tomography (PET) or single photon emission computed tomography (SPECT). One of the most prominent detector materials is cadmium zinc telluride (CZT), which is currently used in several application-specific nuclear imaging systems. In this work, the charge-transport effects in pixelated CZT detectors in relation to detector pixel size and thickness are investigated for pixels sizes from 0.4 up to 1.6 mm. The determination of an optimum pixel size and thickness for use with photon energies of 140 and 511 keV, suitable for SPECT and PET studies, is attempted using photon detection efficiency and energy resolution as figures of merit. The Monte Carlo method combined with detailed finite element analysis was utilized to realistically model photon interactions in the detector and the signal generation process. The GEANT4 Application for Tomographic Emission (GATE) toolkit was used for photon irradiation and interaction simulations. The COMSOL MULTIPHYSICS software application was used to create finite element models of the detector that included charge drift, diffusion, trapping, and generation. Data obtained from the two methods were combined to generate accurate signal induction at the detector pixels. The energy resolution was calculated as the full width at half maximum of the energy spectrum photopeak. Photon detection efficiency was also calculated. The effects of charge transport within the detector and photon escape from primary pixel of interaction were investigated; the extent of diffusion to lateral pixels was also assessed. Charge transport and signal induction were affected by the position of a pixel in the detector. Edge and corner pixels were less susceptible to lateral diffusion than pixels located in the inner part of the detector. Higher detection efficiency and increased photon escape from primary interaction pixel were observed for thicker detectors. Energy resolution achieved better values in 0.7 and 1.0 mm pixel size for 5 mm detector thickness and 1.6 mm pixel size for 10 mm thickness. Selection of pixel size and thickness depends on the imaging application and photon energy utilized. For systems that integrate two nuclear imaging modalities (i.e., combined SPECT/PET), the pixel size should offer an appropriate balance of the effects that take place in the detector, based on the results of the current work. This allows for a system to be designed with the same detector material and the same geometrical configuration for both modalities.

Investigation of a Multi-Anode Microchannel Plate PMT for Time-of-Flight PET

We report on an investigation of a mulit-anode microchannel plate PMT for time-of-flight PET detector modules. The primary advantages of an MCP lie in its excellent timing properties (fast rise time and low transit time spread), compact size, and reasonably large active area, thus making it a good candidate for TOF applications. In addition, the anode can be segmented into an array of collection electrodes with fine pitch to attain good position sensitivity. In this paper, we investigate using the Photonis Planacon MCP-PMT with a pore size of 10 µm to construct a PET detector module, specifically for time-of-flight applications. We measure the single electron response by exciting the Planacon with pulsed laser diode. We also measure the performance of the Planacon as a PET detector by coupling a 4 mm × 4 mm × 10 mm LSO crystal to individual pixel to study its gain uniformity, energy resolution, and timing resolution. The rise time of the Planacon is 440 ps with pulse duration of about 1 ns. A transit time spread of 120 ps FWHM is achieved. The gain is fairly uniform across the central region of the Planacon, but drops off by as much as a factor of 2.5 around the edges. The energy resolution is fairly uniform across the Planacon with an average value of 18.6±0.7% FWHM. While the average timing resolution of 252±7 ps FWHM is achieved in the central region of the Planacon, it degrades to 280±9 ps FWHM for edge pixels and 316±15 ps FWHM for corner pixels. We compare the results with measurements performed with a fast timing conventional PMT (Hamamatsu R-9800). We find that the R9800, which has significantly higher PDE, has a better timing resolution than the Planacon. Furthermore, we perform detector simulations to calculate the improvement that can be achieved with a higher PDE Planacon. The calculation shows that the Planacon can achieve significantly better timing resolution if it can attain the same PDE as the R-9800, while only a 30% improvement is needed to yield a similar timing resolution as the R-9800.

Reconstructing unknown wavefronts by use of diffraction-intensity pattern subtraction

We propose an easy method for unknown wavefront reconstruction by a typical Fourier transform system based on an amplitude-only spatial light modulator (SLM). The Fourier transform of the diffraction intensity pattern on the backfocal plane is the autocorrelation of the wavefront-transmitted SLM. Two intensity patterns were recorded when we set the transmittance of one corner pixel on the SLM as 0 and 1, separately, under the condition that the transmittances of the other pixels equal 1. The complex amplitude of the wavefront-transmitted SLM can be extracted from the Fourier transform of the diffraction intensity pattern subtraction without complicated calculations. The simulation results certifying our theories are also given. Our method provides us with an easy way for detecting the wavefront with high speed and high quality in real time without the reference wave.

Photon Event Centroiding with UV Photon‐counting Detectors

ABSTRACTWe report on photon‐counting data taken with a laboratory prototype detector for the ASTROSAT UV imaging telescope, and the selection of the centroiding algorithms and parameter settings for flight. Thresholds are discussed for event recognition. The optimum event centroid algorithm is a 5 × 5 pixel weighted mean, after bias removal accurate to 2%–3%. This can be achieved within each readout frame by using the lowest value corner pixel of the photon event subarray. We also discuss other corner pixel thresholds for flagging double‐photon events, and methods of correcting centroiding errors with different algorithms and event rates.

Edge effects in pixelated CdZnTe gamma detectors

Edge effects in pixelated CdZnTe detectors are a result of the high dielectric constant of the CdZnTe detector material, and the tendency of the field lines emanating from the hole charge carriers to remain within the detector volume. As a result spectra for edge and corner pixels tend to exhibit a longer low energy tail at the expense of the number of events in the photo-peak. We focus on a thick pixelated CdZnTe detector, where, the detector thickness is comparable to the lateral size. We develop a theoretical Monte-Carlo simulation that well describes the experimentally observed edge effects. We show that when the same detector is embedded in an array of similar detector at the same HV, the edge effects disappear, and the spectral properties of even the central pixels improve noticeably.

Evaluation of CsI(Tl)-PIN diode array tiling schemes

Methods of tiling scintillator-photodiode detector arrays were investigated. Si photodiodes coupled to CsI(Tl) scintillation crystals are being considered for high-resolution (/spl sim/1 mm) gamma-ray imaging. Large area (/spl sim/200 cm/sup 2/) detectors can be assembled by tiling together modular arrays. The goal of the tiling schemes was to maintain uniform pixel pitch and detector performance across the seam where the arrays join. The perimeter of the array has an insensitive area that can be an appreciable fraction of a pixel size when using 1 mm pixels. One tiling scheme used smaller pixels at the edge of the arrays to compensate for this insensitive area (partial PIN method). A second approach used a scintillator light-sharing technique to cover the insensitive area (segmented method). The partial pixel method resulted in worse energy resolution and sensitivity in the partial pixels compared to the full pixels (20% and 30% FWHM at 140 keV for edge and corner pixels, respectively, versus 12% in full pixels, with relative corner pixel sensitivity of /spl sim/50%), but successfully maintained spatial resolution and pixel pitch. The segmented crystal light-sharing method demonstrated 17%-20% FWHM energy resolution (at 140 keV) in the edge pixels, /spl sim/85% relative sensitivity, and provided good pixel identification (i.e., good spatial resolution). However, the segmented crystal method is considerably more complex to implement, compared with the partial pixel method.

Effects of detector scatter on photopeak fraction in pixellated solid state detectors

A Monte Carlo simulation of the photopeak fraction in a 1 cm-thick CdTe crystal with square pixel side dimensions ranging between 0.5 and 2.5 mm was performed. Photons of 140 and 511 keV were assumed to impinge only on a “center” pixel, normal to the crystal surface and uniformly distributed. Three families of pixels were defined relative to the center pixel:(1) the four edge pixels; (2) the four corner pixels; and (3) the 16 next-nearest neighbors. For 140 keV photons, as pixel size is reduced from 2.5 to 0.5 mm, the number of total events in the edge pixels increases from 8% to 17% of the number interacting in the center pixel due to the increasing pixel surface area-to-volume ratio. At 511 keV, however, as the pixel size diminishes the number of events in the edge pixels decreases from 13% to 7% of the number interacting in the center pixel, due to the larger mean free path of the scattered photons and the smaller dimensions of the neighbouring pixels. Although the 1 cm crystal attenuates nearly 99% of the incident 140 keV photons, the computed photofraction for a single pixel ranges from 62% to 75% as pixel size increases from 0.5 to 2.5 mm. For the smallest pixel, 19% and 1.7% of the incident photons result in photoelectric events whose fluorescent photon escapes the pixel for 140 and 511 KeV, respectively. At 511 keV only 42% of incident photons interact with 1 cm CdTe. For the 0.5 mm pixel, the photofraction is only 5.7% at 511 keV. For the 2.5 mm pixel the photofraction improves to 8.3%. These values are roughly half the photofraction of 3/8” NaI at this energy. At 511 keV, of the initially interacting photons, roughly 3/4 of them will Compton scatter to the surrounding pixels and deposit less than photopeak energy. To significantly improve the photopeak performance of pixellated detectors, it appears that simultaneous sampling of neighbouring pixels may be necessary. This study involves the simulation of photon transport within the crystal and has not addressed the effects of electron range, charge diffusion, and induced charge signal on the photofraction.

Identification and measurement of fibers in scanning electron microscopy images using a high-order correlation process

This work describes the development of a dedicated system capable of identifying, measuring and counting various types of fibers and other objects in digitized scanning electron micrograph (SEM) imagery. The system uses a recursive high order correlation (HOC) process to extract the corner pixels of the fibers. The objects are defined by grouping connected corners, so that morphometric analysis can be performed. The method developed performs satisfactorily when the density of fibers per image ranges from low to medium. Simulation results for several cases are presented, along with a discussion on the capabilities and limitations of the current version of the system.

Pseudo one pass thinning algorithm

A One Pass Thinning Algorithm (OPTA) and a Pseudo One Pass Thinning Algorithm (POPTA) are proposed in this paper. The one pass thinning algorithm uses the concept of thinning and restoring. The OPTA deletes pixels from North, South, East, and West in one pass. The POPTA uses an additional concept of corner pixels, which detects the corner pixels and deletes them to yield better skeletons with little or no distortion. Experimental results show that each of the patterns is thinned to a skeleton of unit pixel width, with no loss of connectivity or distortion. The algorithms are width independent and work for both even pixel wide and odd pixel wide images.

Computing a shape's moments from its boundary

Moments have been among the most commonly used tools in the analysis of 2D shapes. Important information about a shape such as its size, center location and orientation are all moment-based attributes. There is also a number of very useful shape features such as maximum and minimum moments of inertia, moment invariants, shape spreadness and shape elongation which are also derived from moments. Traditionally, moments of a shape are computed from the occupancy array representation of the shape. That is, all the shape's pixels are involved in the computation. In this paper we propose a method which computes the same set of moments using only the corner pixels along a shape's boundary. The basic approach is to construct a set of triangles using the shape's corners and the origin of the coordinate system. The moments of these triangles are computed first. The moments of the shape are then derived from the triangles' moments. We have found the suggested method to be much more efficient than the traditional occupancy array based methods.