Abstract
We have developed three semi-automated programs using image geometric transformation which replaces typical works by manual operation.
When performing height measurement or object discrimination using a security video image, the image is distorted by lens aberrations, and it may affect the measurement value. We have developed a program which corrects distorted images, semi-automatically. Distortion corrected images were obtained by the principal component analysis using a known straight line in an image.
Running condition of vehicles in a traffic accident is often estimated by analyzing the tachograph chart. Since the chart is a circular coordinate, reading error may occur. We have developed a program which transforms circle coordinate into rectangle coordinate. Circular tachograph was transformed into rectangular as amplitude vs. angle by polar coordinate transformation.
Perspective transformation of images to the correct position is often done in case of image measurement in order to facilitate subsequent measurement, because various value are conserved in correct position image. Although four precise coordinates were generally required for perspective transformation, we considered it became possible only by a length using characteristics of ellipse. We have developed a program which performed perspective transformation using an elliptic object in an image. The ellipse was fitted using the least squares method by singular value decomposition, and images of correct position were acquired by transformation from an ellipse to a circle.
