KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Settings

Subject: search.filters.subject.Hough transform

Search Results

Now showing 1 - 2 of 2
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Effective horizon detection on complex seas using back propagation neural network
    (2019-07-01)
    Kumeechai, Pisanu
    ;
    Jiriwibhakorn, Somchat
    Object detection is one of the main features of surface vehicles that do not require a driver (USV). Research on the detection of the horizon is one of the basic factors in conducting obstructions. Precisely detection helps to improve the efficiency and accuracy of object detection. Because it will eliminate most irrelevant areas. In this paper, there are four different algorithms that apply to three image sequences that have been taken in the open sea. This paper focuses on the accuracy rate and efficiency of horizon detection. By testing four different algorithms, including Hough transform, least squares, RANSAC and neural networks to separate the horizon from the image. Experimental results show that artificial neural network methods use more time but accuracy better than others.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Image registration using Hough transform and phase correlation
    (2006-01-01)
    Chunhavittayatera, Siwaphon
    ;
    Chitsobhuk, Orachat
    ;
    Tongprasert, Kiatnarong
    This paper presents image registration using a combination of Hough transform and phase correlation technique in Fourier domain to estimate rotation and translation parameters. These paramters are used to register input images and create a seamless representation of the registered image. The first step is based on 1D phase correlation which calculates the translation of angle in Hough space, obtained form Hough transform. The rotation parameter obtain from this step is used to de-rotated the input images. The second step provides the translation in x-y axis using 2D phase correlation. Finally, the input images are registered using the estimated translation parameter. The experimental results show the accuracy of the proposed method to detect the translation parameters and find the relevant potential angles of rotation for various image details and sizes. The proposed method is ease to operation, less computation complexity, thus requires less amount of process time due to the high efficiency of fast Fourier transform (FFT).