Browsing by Author "Ikokou, Guy Blanchard"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Open Access
    A technique for optimal selection of segmentation scale parameters for object-oriented classification of urban scenes
    (2013) Ikokou, Guy Blanchard; Smit, Julian
    Multi-scale image segmentation produces high level object features at more than one level, compared to single scale segmentation. Objects generated from this type of segmentation hold additional attributes such as mean values per spectral band, distances to neighbouring objects, size, and texture, as well as shape characteristics. However, the accuracy of these high level features depends on the choice of segmentation scale parameters. Several studies have investigated techniques for scale parameter selection. These proposed approaches do not consider the different objects’ size variability found in complex scenes such as urban scene as they rely upon arbitrary object size measures, introducing instability errors when computing image variances. A technique to select optimal segmentation scale parameters based on image variance and spatial autocorrelation is presented in this paper. Optimal scales satisfy simultaneously the conditions of low object internal variance and high inter-segments spatial autocorrelation. Applied on three Cape Town urban scenes, the technique produced visually promising results that would improve object extraction over urban areas.
  • No Thumbnail Available
    Item
    Open Access
    Analytical camera pose estimation and inverse modelling of high order radial lens distortion polynomials for close range photogrammetry
    (2025) Ikokou, Guy Blanchard; Shoko, Moreblessings; Tagoe, Naa Dedei
    Camera calibration aims at estimating intrinsic and extrinsic camera parameters that accurately describe the projection of points from the 3D scene to the 2D image sensor. The absence of numerical estimates of parameters describing the camera internal and external geometry can prevent absolute readjustment of bundles of rays that project 3D points from the scene onto the image plane. When information derived from photographs is used for metric purposes, small imaging errors can significantly affect the accuracy of derived information. Current analytic a l method for camera pose calibration failed to exploit the intrinsic geometric properties associated with each camera parameter within the structures of individual coefficients of the projective transformation matrix. This resulted in these methods being prone to parameter coupling, sign ambiguity, and multiple roots associated with some of the parameters. In the same way, recursive reversions methods proposed to compute inverses of radial distortion coefficients in order to correct radial distortions were found not suitable for high-degree radial distortion polynomials and failed to model the inverse profiles of severe barrel, pincushio n, and moustache distortions inherent to consumer-grade cameras used in close range photogrammetry. This PhD research successfully developed two analytical calibration systems. The first calibration system decomposed the coefficients of the projective transformation matrix into nineteen robust equations that independently estimate individual parameters describing the internal geometry of the camera as well as its location and orientation in a 3D scene. The second calibration system, based on the concept of forced differential equation, successfully modelled inverse coefficients of high-order quintic, sextic, and octic radial distortion polynomials and reduced the effects of severe barrel, pincushion, and moustache radial distortions on distorted points. The experimental results demonstrate that both developed calibration strategies achieved root mean square reprojection errors of 0.0843 and 0.057 pixels, respectively. These values are substantially lower than those reported by several current state- of-the-art methods, which exhibit average errors around 3.63 pixels. Such a signific a nt reduction in error by nearly two orders of magnitude not only confirms the high accuracy of the proposed approaches but also underscores their suitability for close-range photogramme tr y applications where sub-pixel precision is essential. In practical terms, this level of accuracy enables more reliable 3D reconstructions and measurements in scenarios where even small deviations can lead to substantial errors. These results affirm the reliability and effective ness of the proposed calibration strategies, making them well-suited for high-precision 3D reconstruction, industrial metrology, cultural heritage documentation, and other close-range imaging tasks where geometric accuracy is paramount.
  • Thumbnail Image
    Item
    Open Access
    Urban scene description for a multi scale classication of high resolution imagery case of Cape Town urban Scene
    (2012) Ikokou, Guy Blanchard; Sithole, George
    In this paper, a multi level contextual classification approach of the City of Cape Town, South Africa is presented. The methodology developed to identify the different objects using the multi level contextual technique comprised three important phases.