Research on Feature Extraction and Segmentation of Rover Wheel Imprint

Nan Li,Hai-bo Gao,Liang Ding,Feng-tian Lv,Zhong-yan Bi,Yi-da Wang

DOI: https://doi.org/10.1007/s11227-018-2502-7

2018-01-01

Abstract:The wheel imprint photograph of the rover contains important information such as wheel motion parameters, soil characteristic parameters and movement status of the rover. Segmenting the wheel imprint area from the photograph is an important prerequisite for feature extraction and parameter identification. Traditional image segmentation cannot take both speed and precision into account because they are based on a wide range of image object attributes, such as grayscale threshold, color, texture, gradient, contrast, shape and size. This paper presents an image segmentation method based on the wheel imprint feature. Compared to other common graphics segmentation methods, the morphological characteristics and frequency domain characteristics of the wheel imprint are found out by analyzing the mechanism of the wheel–terrain interaction process, and the eigenvector is also constructed. In the wheel trace feature space, the clustering algorithm is used to divide the image into the trace area and the non-trace area. The segmentation accuracy and processing speed were used to evaluate the algorithm. The experimental results show that the developed algorithm based on the characteristics of wheel imprint formation mechanism is more accurate and efficient than the traditional method.