Automobile Longitudinal Axis Detection Method Based on Image Segmentation and Preliminary Results

With the growing number and high usage frequency, it was important for automobiles to test the performance. In order to detect automobiles effectively, the technologies of automobile contour detection and longitudinal axis extraction, based on digital image processing, were studied. The basic concepts of automobile testing technology were introduced, and several commonly used image segmentation methods were analyzed. Before image segmentation, the automobile image was preprocessed, including gray scale transformation, gray scale stretching and median filtering. According to the monotonicity of interclass variances to both sides of threshold, the rapid realization method of image segmentation based on OTSU was proposed. The extraction of automobile longitudinal axis was realized by using approximation method. The software running showed that it could effectively detect automobile contour and extract longitudinal axis, which laid foundation for subsequent automobile image analysis and feature extraction, and it had certain practical value.