Hou, G.-J.; Luan, X.; Song, D.-L., and Ma, X.-Y., 2016. Underwater mad-made object recognition on the basis of color and shape features.

In complex underwater situations, how to realize object extraction accurately and effectively is the key technology of underwater object recognition. In this paper, the detection and recognition techniques of underwater man-made objects on the basis of color and shape features have been studied in depth. First, the objects of interest in an underwater image are extracted by applying a color-based algorithm. Then an improved two-dimensional Otsu algorithm is utilized for removing the background color noise. To recognize the shape type of a regular object, a robust algorithm based on shape signature is presented. The experimental results show that the proposed approach is effective and robust, such as an acceptable extraction rate (exceeding 80%) of the object of interest, an ideal outcome of background color noise removal, high accurate shape of the object's edge, and a good average recognition rate of shape type (approximately 90%). It proves that this algorithm can accurately settle the problem of object extraction and recognition under different cases of distance, angle, and illumination.