Massey Documents by Type

Permanent URI for this communityhttps://mro.massey.ac.nz/handle/10179/294

Browse

Subject: search.filters.subject.Imaging systems

Search Results

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    Explainable spectral super-resolution based on a single RGB image : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy (PhD) in Electronics and Computer Engineering at Massey University, Manawatu, New Zealand
    (Massey University, 2024-04-09) Chang, Yuan
    Hyperspectral imaging offers fine spectral measurements of target surfaces, finding utility in various fields. However, traditional hyperspectral systems grapple with high-cost issues. On the other hand, conventional RGB cameras, which provide relatively coarse measurements of surface spectra, are widely accessible. Consequently, the recovery of spectral information from RGB images has emerged as a popular approach for low-cost hyperspectral imaging, a venture also known as single-image spectral super-resolution. Yet, existing methods, mostly rooted in deep convolutional neural networks, tend to suffer from limited interpretability. In our research, we propose an explainable method for single-image spectral super-resolution. This method relies on the RGBPQR colour space, a low-dimensional spectral data model representing the spectrum. Leveraging the RGBPQR spectral model, we can transform the spectral reconstruction task into a regression problem. To tackle the metamerism issue, we analysed existing spectral super-resolution networks and discovered that these networks often depend on local textural information as context to mitigate metamerism. Informed by this insight, we utilized features extracted from multiscale local binary patterns as contextual information to design our explainable method. Furthermore, in this study, we discussed the error measurements and loss functions employed in this research area and proposed a new error measurement that can represent performance more accurately. We also endeavoured to put forward a method for quantitatively measuring the ability to resolve metamerism, a critical problem in spectral super-resolution. Through our research, we offered a simple, low-dimensional, and explainable spectral super-resolution solution.
  • Thumbnail Image
    Item
    A portable multi-modal micro-imaging system for automated scanning and image stitching applications : a thesis submitted to Massey University in accordance with the requirements of the degree of Master of Engineering in the School of Engineering and Advanced Technology
    (Massey University, 2019) Naqvi, Adam
    Microscopic imaging is an important element in many fields like biology, medicine, diagnostics, engineering, and materials research. Muti-modal microscopes are ideal for imaging samples that reveal unseen structures that could not otherwise be seen with normal bright-field microscopes. Point-of-care micro-imaging devices are ones that can deliver the features of a microscope in areas where access to a laboratory or medical facilities are scarce. This thesis presents the development of a portable micro-imaging system that uses multi-modal illumination to image samples in bright-field, fluorescence, ambient and laser diffraction modes. A systematic design method has been used to develop the system from the conceptual phase to a working prototype. The system incorporates variable magnification through an inverted turret system and a GUI application for live image view, automatic scanning, auto-focusing and image processing. The utility of the system is demonstrated through imaging stained biological samples for a local industry application. The acquired images are measured against sharpness and noise. It is observed that the sharpness and noise of the images produced vary with the type of sample: samples with higher contrast generally produce sharper images with less noise. It has also been found that diffused ambient illumination produces the most consistent sharpness and noise scores between magnifications. Performance of algorithms used is discussed and improvements are suggested for building a more compact and stable platform including a method to calibrate measurements for particle size estimation.
  • Thumbnail Image
    Item
    Investigation of low resolution point clouds for illumination correction in pushbroom hyperspectral images : a thesis presented in partial fulfilment of the requirements for the degree of Master of Engineering in Machatronics at Massey University, Turitea Campus, Palmerston North, New Zealand
    (Massey University, 2018) Haarhoff, William B.
    Global food demand is predicted to double between 2015 and 2050. Current agricultural production is unable to facilitate this growth. Consequently, plant breeding must be accelerated to breed improved cultivars that can meet this demand. While technologies such as genomics are suitable for accelerating plant breeding, phenotyping lags behind and is currently considered the bottleneck. Consequently, imaging and remote sensing technologies are being used to provide quantitative, reliable phenotype information. One such technology; hyperspectral imaging can provide physiological, biophysical, and biochemical phenotypic information. While hyperspectral imaging has reached a substantial level of maturity in aerial and satellite based remote sensing applications, it is still underdeveloped in the close-range lab-based phenotyping scenario. In particular is the effect of illumination and complex plant geometry which affects the measured signal and is even more pronounced in the close range hyperspectral imaging. Methods for correction of illumination/geometry effects developed for aerial, and satellite-based imaging are unsuitable for close range hyperspectral imaging. Recently there has been an interest in fusing hyperspectral images with point clouds captured by 3D imaging devices to provide more comprehensive high dimensional phenotype information. However, one study focusses on the possibility of using 3D geometry of the plant to correct for the effects of illumination in hyperspectral images. This study investigates the use of low resolution point clouds captured with low cost devices for use in illumination modelling and correction of hyperspectral images acquired in close range lab-based scenario.
  • Thumbnail Image
    Item
    Automated body condition scoring of dairy cattle : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Engineering at Massey University, Manawatū, New Zealand
    (Massey University, 2018) O'Connor, Aaron
    This research demonstrates the development and implementation of an automatic body condition scoring system for dairy cattle that can operate in a real-world environment. Body condition scoring is a subjective method used for measuring changes in energy reserves in many animals, including dairy cattle. These energy reserves can be measured by analysing specific regions on the cow to estimate the amount of fat the animal is carrying. This information allows for greater management of the herd by adjusting the feeding strategies to ensure that each cow is at an optimal condition score. Maintaining an optimal condition throughout the year has implications for milk yield, reproductive performance, animal welfare, and overall farm profits. Current condition scoring methods are manual and are highly subjective, time consuming, expensive, and require a high level of training and competency. These limitations have created a demand for an accurate and objective scoring system. This research presents an automated system that utilises a single camera to be placed above the path of the cow at the entrance or exit to a milking platform or weigh scale. When the cow passes in view of the camera, the features are automatically extracted and converted to a conditions score. Tests have shown that the system successfully predicted the condition score within half a point of the true score for 83% of the 710 cows scored, and 96% within one point.
  • Thumbnail Image
    Item
    Development of an in-field tree imaging system : a thesis presented in partial fulfilment of the requirements for the degree of Master of Technology at Massey University
    (Massey University, 1996) Weehuizen, Marijn
    Quality inventory information is essential for optimal resource utilisation in the forestry industry. In-field tree imaging is a method which has been proposed to improve the preharvest inventor assessment of standing trees. It involves the application of digital imaging technology to this task. The method described generates a three dimensional model of each tree through the capture of two orthogonal images from ground level. The images are captured and analysed using the "TreeScan" in-field tree imaging system. This thesis describes the design, development, and evaluation of the TreeScan system. The thesis can also be used as a technical reference for the system and as such contains appropriate technical and design detail. The TreeScan system consists of a portable computer, a custom designed high resolution scanner with integral microcontroller, a calibration rod, and custom designed processing software. Images of trees are captured using the scanner which contains a CCD line scan camera and a precision scanning mechanism. Captured images are analysed on the portable computer using customised image processing software to estimate real world tree dimensions and shape. The TreeScan system provides quantitative estimates of five tree parameters; height, sweep, stem diameter, branch diameter, and feature separation such as internodal distance. In addition to these estimates a three dimensional model is generated which can be further processed to determine the optimal stem breakdown into logs.