Polarimetric imaging data contains rich information about surface textures, shape, and shading of objects in a scene, which can be used to discriminate objects from background. Due to spectral signatures are limited to material properties, separating man-made objects from natural scene is a difficult task in complex scene. In this paper, we present a man-made object separation technique from natural scene utilizing polarimetric data. We started by calculating different polarimetric component images based on Stokes vector measurement. After that, combining polarimetric component images into color space and single-channel conversion, a polarimetric signature image is calculated. Then, considering pixel neighborhood relationship, an incremental clustering approach is applied to group similar pixel patterns of polarimetric signature image. Finally, a morphological structuring element is applied to conduct the morphological close operation to refine the final background mask. Ground truth is generated manually. A high-performing score (Dice Similarity Coefficient (DSC)) is achieved on the final man-made object area mask which separates man-made objects well from natural scene. Future work will exploit the use of multispectral polarimetric imagery for target classification using machine learning techniques.
KEYWORDS: Point spread functions, Single photon emission computed tomography, Modulation transfer functions, Spatial resolution, Monte Carlo methods, Reconstruction algorithms, Collimators, Imaging systems, Fourier transforms
Light field SPECT (L-SPECT) is an improved version of SPECT and works by introducing the concept of plenoptic imaging to reduce scanning time and to increase the amount of detected information. In L-SPECT, a tungsten pinhole array is used as a collimator to differentiate the incoming direction of radiation, rather than only allowing radiation from a set direction dictated by a conventional tube collimator. The distance of the pinhole array to the sensors’ plane is so that the sensors behind each pinhole are only exposed through that pinhole alone. This paper investigates the effects of the pinholes’ diameter and pitch over the reconstruction resolution using simulation experiments. In this proposed reconstruction algorithm, a ray is back projected from the centre of each detector with non-zero pixel value via the corresponding pinhole’s centre, and towards the area of interest with 128×128×128 voxels. The projected rays’ intersections are identified by using ray tracing and the voxels at which they intersect are updated by incrementing with the sum of the pixel values from each detector involved. Experiments are conducted with pinhole arrays of 100×100, 50×50, 30×30 and pinhole diameter of 0.5mm, 1mm and 2mm. Reconstruction is conducted for various simulated objects. Results indicate that when the number of pinholes is increased, the diameter of the pinholes should be reduced to maintain spatial resolution. Moreover, a reconstruction performed by using only 12 projections shows similar quality for the same with 36 and 72 projections. The analysis of the proposed reconstruction algorithm shows that it improves spatial resolution over the filtered back projection algorithm. Reconstruction quality can be further improved by considering scattering loss and photon attenuation.
Image registration is a fundamental image processing technique. It is used to spatially align two or more images that have been captured at different times, from different sensors, or from different viewpoints. There have been many algorithms proposed for this task. The most common of these being the well-known Lucas–Kanade (LK) and Horn–Schunck approaches. However, the main limitation of these approaches is the computational complexity required to implement the large number of iterations necessary for successful alignment of the images. Previously, a multi-pass image interpolation algorithm (MP-I2A) was developed to considerably reduce the number of iterations required for successful registration compared with the LK algorithm. This paper develops a kernel-warping algorithm (KWA), a modified version of the MP-I2A, which requires fewer iterations to successfully register two images and less memory space for the field-programmable gate array (FPGA) implementation than the MP-I2A. These reductions increase feasibility of the implementation of the proposed algorithm on FPGAs with very limited memory space and other hardware resources. A two-FPGA system rather than single FPGA system is successfully developed to implement the KWA in order to compensate insufficiency of hardware resources supported by one FPGA, and increase parallel processing ability and scalability of the system.
3D motion capture is difficult when the capturing is performed in an outdoor environment without controlled surroundings. In this paper, we propose a new approach of using two ordinary cameras arranged in a special stereoscopic configuration and passive markers on a subject’s body to reconstruct the motion of the subject. Firstly for each frame of the video, an adaptive thresholding algorithm is applied for extracting the markers on the subject’s body. Once the markers are extracted, an algorithm for matching corresponding markers in each frame is applied. Zhang’s planar calibration method is used to calibrate the two cameras. As the cameras use the fisheye lens, they cannot be well estimated using a pinhole camera model which makes it difficult to estimate the depth information. In this work, to restore the 3D coordinates we use a unique calibration method for fisheye lenses. The accuracy of the 3D coordinate reconstruction is evaluated by comparing with results from a commercially available Vicon motion capture system.
Whiplash-associated disorder (WAD) is a commonly occurring injury that often results from neck trauma suffered in car accidents. However the cause of the condition is still unknown and there is no definitive clinical test for the presence of the condition. Researchers have begun to analyze the size of neck muscles and the presence of fatty infiltrates to help understand WAD. However this analysis requires a high precision delineation of neck muscles which is very challenging due to a lack of distinctive features in neck magnetic resonance imaging (MRI). This paper presents a novel atlas-based neck muscle segmentation method which employs discrete cosine-based elastic registration with affine initialization. Our algorithm shows promising results based on clinical data with an average Dice similarity coefficient (DSC) of 0.84±0.0004.
This paper aims to investigate the performance of a newly proposed L-SPECT system for small animal brain imaging. The L-SPECT system consists of an array of 100 × 100 micro range diameter pinholes. The proposed detector module has a 48 mm by 48 mm active area and the system is based on a pixelated array of NaI crystals (10×10×10 mm elements) coupled with an array of position sensitive photomultiplier tubes (PSPMTs). The performance of this system was evaluated with pinhole radii of 50 μm, 60 μm and 100 μm. Monte Carlo simulation studies using the Geant4 Application for Tomographic Emission (GATE) software package validate the performance of this novel dual head L-SPECT system where a geometric mouse phantom is used to investigate its performance. All SPECT data were obtained using 120 projection views from 0° to 360° with a 3° step. Slices were reconstructed using conventional filtered back projection (FBP) algorithm. We have evaluated the quality of the images in terms of spatial resolution (FWHM) based on line spread function, the system sensitivity, the point source response function and the image quality. The sensitivity of our newly proposed L- SPECT system was about 4500 cps/μCi at 6 cm along with excellent full width at half-maximum (FWHM) using 50 μm pinhole aperture at several radii of rotation. The analysis results show the combination of excellent spatial resolution and high detection efficiency over an energy range between 20-160 keV. The results demonstrate that SPECT imaging using a pixelated L-SPECT detector module is applicable in a quantitative study of mouse brain imaging.
Radiolabeled tracer distribution imaging of gamma rays using pinhole collimation is considered promising for small animal imaging. The recent availability of various radiolabeled tracers has enhanced the field of diagnostic study and is simultaneously creating demand for high resolution imaging devices. This paper presents analyses to represent the optimized parameters of a high performance pinhole array detector module using two different characteristics phantoms. Monte Carlo simulations using the Geant4 application for tomographic emission (GATE) were executed to assess the performance of a four head SPECT system incorporated with pinhole array collimators. The system is based on a pixelated array of NaI(Tl) crystals coupled to an array of position sensitive photomultiplier tubes (PSPMTs). The detector module was simulated to have 48 mm by 48 mm active area along with different pinhole apertures on a tungsten plate. The performance of this system has been evaluated using a uniform shape cylindrical water phantom along with NEMA NU-4 image quality (IQ) phantom filled with 99mTc labeled radiotracers. SPECT images were reconstructed where activity distribution is expected to be well visualized. This system offers the combination of an excellent intrinsic spatial resolution, good sensitivity and signal-to-noise ratio along with high detection efficiency over an energy range between 20-160 keV. Increasing number of heads in a stationary system configuration offers increased sensitivity at a spatial resolution similar to that obtained with the current SPECT system design with four heads.
This paper focuses on minimizing the time requirement for CT capture through an innovative simultaneous X-ray capture method. The concept was presented in previous publications with synthetically sampled data from a synthetic phantom. This paper puts emphasis on real data reconstruction where a physical 3D phantom consisting of simple geometric shapes was used for the experiment. For a successful reconstruction of the physical phantom, precise calibration of the setup is ensured in this work. Targeting better reconstruction from minimal number of iterations, the sparsity prior CT reconstruction algorithm proposed by Saha et al. [11]was adapted to work in conjunction with the simultaneous X-ray capture modality. Along with critical evaluations of the experimental findings, this paper focuses on optimal parameter settings to achieve a given reconstruction resolution.
KEYWORDS: 3D image processing, Ultrasonography, 3D acquisition, Information operations, Compressed sensing, Stereoscopy, Wavelets, 3D image reconstruction, Data processing, Wavelet transforms
Ultrasound (US) imaging is one of the most popular medical imaging modalities, with 3D US imaging gaining popularity recently due to its considerable advantages over 2D US imaging. However, as it is limited by long acquisition times and the huge amount of data processing it requires, methods for reducing these factors have attracted considerable research interest. Compressed sensing (CS) is one of the best candidates for accelerating the acquisition rate and reducing the data processing time without degrading image quality. However, CS is prone to introduce noise-like artefacts due to random under-sampling. To address this issue, we propose a combined prior-based reconstruction method for 3D US imaging. A Laplacian mixture model (LMM) constraint in the wavelet domain is combined with a total variation (TV) constraint to create a new regularization regularization prior. An experimental evaluation conducted to validate our method using synthetic 3D US images shows that it performs better than other approaches in terms of both qualitative and quantitative measures.
Motion estimation or optic flow computation for automatic navigation and obstacle avoidance programs running on Unmanned Aerial Vehicles (UAVs) is a challenging task. These challenges come from the requirements of real-time processing speed and small light-weight image processing hardware with very limited resources (especially memory space) embedded on the UAVs. Solutions towards both simplifying computation and saving hardware resources have recently received much interest. This paper presents an approach for image registration using binary images which addresses these two requirements. This approach uses translational information between two corresponding patches of binary images to estimate global motion. These low bit-resolution images require a very small amount of memory space to store them and allow simple logic operations such as XOR and AND to be used instead of more complex computations such as subtractions and multiplications.
3D ultrasound imaging has advantages as a non-invasive and a faster examination procedure capable of displaying volume information in real time. However, its resolution is affected by speckle noise. Speckle reduction and feature preservation are seemingly opposing goals. In this paper, a nonlinear multi-scale complex wavelet diffusion based algorithm for 3D ultrasound imaging is introduced. Speckle is suppressed and sharp edges are preserved by applying iterative multi-scale diffusion on the complex wavelet coefficients. The proposed method is validated using synthetic, real phantom, and clinical 3D images, and it is found to outperform other methods in both qualitative and quantitative measures.
The purpose of this study is to derive optimized parameters for a detector module employing an off-the-shelf X-ray
camera and a pinhole array collimator applicable for a range of different SPECT systems. Monte Carlo simulations using
the Geant4 application for tomographic emission (GATE) were performed to estimate the performance of the pinhole
array collimators and were compared to that of low energy high resolution (LEHR) parallel-hole collimator in a four head
SPECT system. A detector module was simulated to have 48 mm by 48 mm active area along with 1mm, 1.6mm and 2
mm pinhole aperture sizes at 0.48 mm pitch on a tungsten plate. Perpendicular lead septa were employed to verify
overlapping and non-overlapping projections against a proper acceptance angle without lead septa. A uniform shape
cylindrical water phantom was used to evaluate the performance of the proposed four head SPECT system of the pinhole
array detector module. For each head, 100 pinhole configurations were evaluated based on sensitivity and detection
efficiency for 140 keV ɣ-rays, and compared to LEHR parallel-hole collimator. SPECT images were reconstructed based
on filtered back projection (FBP) algorithm where neither scatter nor attenuation corrections were performed. A better
reconstruction algorithm development for this specific system is in progress. Nevertheless, activity distribution was well
visualized using the backprojection algorithm. In this study, we have evaluated several quantitative and comparative
analyses for a pinhole array imaging system providing high detection efficiency and better system sensitivity over a large
FOV, comparing to the conventional four head SPECT system. The proposed detector module is expected to provide
improved performance in various SPECT imaging.
Existing Computed Tomography (CT) systems require full 360 rotation projections. Using the principles of lightfield
imaging, only 4 projections under ideal conditions can be sufficient when the object is illuminated with multiple-point Xray
sources. The concept was presented in a previous work with synthetically sampled data from a synthetic phantom.
Application to real data requires precise calibration of the physical set up. This current work presents the calibration
procedures along with experimental findings for the reconstruction of a physical 3D phantom consisting of simple
geometric shapes. The crucial part of this process is to determine the effective distances of the X-ray paths, which are not
possible or very difficult by direct measurements. Instead, they are calculated by tracking the positions of fiducial
markers under prescribed source and object movements. Iterative algorithms are used for the reconstruction. Customized
backprojection is used to ensure better initial guess for the iterative algorithms to start with.
Ultrasound imaging is a dominant tool for diagnosis and evaluation in medical imaging systems. However, as its major limitation is that the images it produces suffer from low quality due to the presence of speckle noise, to provide better clinical diagnoses, reducing this noise is essential. The key purpose of a speckle reduction algorithm is to obtain a speckle-free high-quality image whilst preserving important anatomical features, such as sharp edges. As this can be better achieved using multiple ultrasound images rather than a single image, we introduce a complex wavelet-based algorithm for the speckle reduction and sharp edge preservation of two-dimensional (2D) ultrasound images using multiple ultrasound images. The proposed algorithm does not rely on straightforward averaging of multiple images but, rather, in each scale, overlapped wavelet detail coefficients are weighted using dynamic threshold values and then reconstructed by averaging. Validation of the proposed algorithm is carried out using simulated and real images with synthetic speckle noise and phantom data consisting of multiple ultrasound images, with the experimental results demonstrating that speckle noise is significantly reduced whilst sharp edges without discernible distortions are preserved. The proposed approach performs better both qualitatively and quantitatively than previous existing approaches.
n this paper, we present an innovative iterative algorithm for tomographic reconstruction. Algebraic reconstruction technique (ART) which is considered as the core of iterative approach has been enhanced to ensure much finer and faster reconstruction. Backprojection has been customized to make it work even when the projections are not uniformly distributed. Contour information of the object has been combined with customized backprojection to ensure a better initial guess to start ART iterations. Based on experiments with both simulated and real medical images it has been shown that the proposed modality is capable of computing more accurate reconstructions in addition with lower computational cost than traditional ART.
Existing Computed Tomography (CT) systems are vulnerable to internal organ movements. This drawback is
compensated by extra exposures and digital processing. CT being a radiation dose intensive modality, it is imperative to
limit the patient’s exposure to X-ray radiation, if only by removing the necessity to take extra exposures. A multiple
pinhole camera, akin to optical lightfield imaging, to acquire simultaneously multiple X-ray projections is presented.
This new method allows a single snapshot acquisition of all necessary projections for 3D reconstruction. It will also
allow the real-time dynamic 3D X-ray reconstruction of moving organs, as it requires no scanning and no moving parts
in its final implementation. A proof-of-concept apparatus that simulates the intended process was built and parallaxed
images were obtained with minor processing. Synthetic 3D reconstruction tests are also presented.
This paper focuses on tomographic reconstruction from a smaller number of projections than usual. Whereas traditional
CT scanner are based on sequential X-ray sources, the proposed methodology in this work is based on simultaneous x-ray
sources on each projection. Simulations have shown that only four projections are needed to reconstruct a slice,
which are captured simultaneously, offering drastic reduction of image capture time. Algebraic Reconstruction
Technique (ART) has been used for reconstruction. Although ART has many advantages over the established
methods, it remained unpopular due to its high computational cost, and most importantly due to the artefacts caused by
the patient's movement during image capture. The simultaneity of the projections helps to overcome this serious
shortcoming of ART.
There exist limitations in the human visual system (HVS) which allow images and video to be reconstructed using fewer
bits for the same perceived image quality. In this paper we will review the basis of spatial masking at edges and show a
new method for generating a just-noticeable distortion (JND) threshold. This JND threshold is then used in a spatial
noise shaping algorithm using a compressive sensing technique to provide a perceptual coding approach for JPEG2000
coding of images. Results of subjective tests show that the new spatial noise shaping framework can provide significant
savings in bit-rate compared to the standard approach. The algorithm also allows much more precise control of distortion
than existing spatial domain techniques and is fully compliant with part 1 of the JPEG2000 standard.
It is often useful to fuse remotely sensed data taken from different sensors. However, before this multi-sensor data fusion
can be performed the data must first be registered. In this paper we investigate the use of a new information-theoretic
similarity measure known as Cross-Cumulative Residual Entropy (CCRE) for multi-sensor registration of remote sensing
imagery. The results of our experiments show that the CCRE registration algorithm was able to automatically register
images captured with SAR and optical sensors with 100% success rate for initial maximum registration errors of up to 30
pixels and required at most 80 iterations in the successful cases. These results demonstrate a significant improvement
over a recent mutual-information based technique.
The detection of objects from a cluttered background using remote sensing data may cause many false alarms if the
target object and the background have overlapping spectra. In this study, we propose an integrated approach to combine
pixel-based spectral labeling with object-based spatial property measures. A hierarchical structure is developed in which
multileveled attributions and decision rules can be implemented. The targets are then extracted progressively.
Experimental results show a substantial reduction in the number of false alarms with the proposed method.
For multi-sensor registration, previous techniques typically use mutual information (MI) rather than the sum-of-the-squared
difference (SSD) as the similarity measure. However, the optimization of MI is much less straightforward than is
the case for SSD-based algorithms. A new technique for image registration has recently been proposed that uses an
information theoretic measure called the Cross-Cumulative Residual Entropy (CCRE). In this paper we show that using
CCRE for multi-sensor registration of remote sensing imagery provides an optimization strategy that converges to a
global maximum with significantly less iterations than existing techniques and is much less sensitive to the initial
geometric disparity between the two images to be registered.
In this paper we introduce and test a new similarity measure for use in a template matching process for target detection
and recognition. The measure has recently been developed for multi-modal registration of medical images and is known
as phase mutual information (PMI). The key advantage of PMI is that it is invariant to lighting conditions, the ratio
between foreground and background intensity and the level of background clutter, which is critical for target detection
and recognition from the surveillance images acquired from various sensors. Several experiments were conducted using
real and synthetic datasets to evaluate the performance of PMI when compared with a number of commonly used
similarity measures including mean squared difference, gradient error and intensity mutual information. Our results show
that PMI consistently provided the most accurate detection and recognition performance.
Registration of two dimensional to three dimensional orthopaedic medical image data has important applications
particularly in the area of image guided surgery and sports medicine. Fluoroscopy to computer tomography (CT)
registration is an important case, wherein digitally reconstructed radiographs derived from the CT data are registered to
the fluoroscopy data. Traditional registration metrics such as intensity-based mutual information (MI) typically work
well but often suffer from gross misregistration errors when the image to be registered contains a partial view of the
anatomy visible in the target image. Phase-based MI provides a robust alternative similarity measure which, in addition
to possessing the general robustness and noise immunity that MI provides, also employs local phase information in the
registration process which makes it less susceptible to the aforementioned errors. In this paper, we propose using the
complex wavelet transform for computing image phase information and incorporating that into a phase-based MI
measure for image registration. Tests on a CT volume and 6 fluoroscopy images of the knee are presented. The femur
and the tibia in the CT volume were individually registered to the fluoroscopy images using intensity-based MI,
gradient-based MI and phase-based MI. Errors in the coordinates of fiducials present in the bone structures were used to
assess the accuracy of the different registration schemes. Quantitative results demonstrate that the performance of
intensity-based MI was the worst. Gradient-based MI performed slightly better, while phase-based MI results were the
best consistently producing the lowest errors.
Background subtraction is widely used for the detection of objects moving against a background in video. Moving objects are detected by comparing the current image with the extracted background image (BI). The BI is constructed using a single or several frames without moving foreground object in the scene. The pixel in BI is updated by the current pixel value if this pixel is determined to be part of the background during the subtraction process. The drawback of this approach is the requirement of the establishment of a background image prior to the detection; otherwise any object that appears in the first frame is detected as “moving object” through the whole sequence. An even more serious problem occurs when there is a sudden change in the background, such as a light being turned on or off, or a newly arrived “still” object. As long as the pixel value change is larger than the threshold, the “still” object after the sudden change will not be included in the background image and hence it will appear as a moving object in the following frames. To avoid these problems, we propose an approach in which a second updated background image, BI2, is stored. BI2 is initially constructed from the first frame detected and then updated through the detection processes with criteria different to that used in updating BI. The pixels in BI2 are updated if they have been determined as background pixels by comparing the difference between the current and previous frames. By using this method, the “still” objects are not falsely detected as moving objects. After a few frames, the “still” objects are updated into the background image BI2. BI2 is then incorporated into the background image BI. Moving objects are then subtracted from the modified background image BI and the “still” objects are eliminated.
In this paper, we present a multisensor surveillance system that consists of an optical sensor and an infrared sensor. In this system, a background subtraction method based on the zero-order statistics is presented for the moving object segmentation. Additionally, we propose an iterative method for multisensor video registration based on the robust Hausdorff distance. An efficient face detection system is shown as an application that will have enhanced performance based on the registration and fusion of the information from the two sensors. Experimental results show the efficacy of the proposed system.
It is well known that tile boundary artefacts occur in lossy wavelet-based image coding. The base model of the JPEG2000 standard (ie JPEG2000 Part I) suffers from these artefacts, being a wavelet-based coding system. This paper analyses the tile boundary problems of JPEG2000 Part I and presents a novel method for reducing these tile boundary artefacts. This method has recently been adopted as part of the JPEG 2000 Verification Model 9.0 and as an addition to Part II of the JPEG2000 standard.
The traditional manual method of path profile analysis is very time consuming and takes approximately 20 - 30 minutes for each path. With the advent of digital terrain data and high-speed computing, this process can be readily automated and the time for each path reduced to a fraction of a second. The calculation of point-to-point losses, however, still presents the user with limited information, and does not readily advise the planner of the coverage of a given transmitter. In this paper, we present a new approach based on area planning.
The wavelet transform is usually performed on a whole image. However when the amount of memory available for the transformation is limited, the input image is partitioned into non-overlapping blocks and then each block is processed independently. Quantization, that typically follows the transformation procedure in a compression system, inevitably introduces distortion, which becomes especially pronounced along the boundaries of the blocks. In this paper we show that a significant reduction in these block boundary artifacts can be achieved by choosing odd block sizes given by 2N + 1 rather than the conventional even block sizes given by 2N. We show that, for the same coefficient entropy, an image compressed using 17 X 17 blocks of wavelet coefficients has at least 1 dB higher PSNR than an image compressed using 16 X 16 blocks of wavelet coefficients.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.