This paper describes a novel approach to object detection from sidescan sonar (SSS) acoustical images. The current techniques of acoustical images processing consume a great deal of time and computational resources with many parameters to tune in order to obtain good quality images. This is due to the handling of the large data volume generated by these kinds of devices. The technique proposed in this work does not make any a priori assumption about the nature of the SSS image to be processed. However, it is able to make a segmentation of the image into two types of regions: acoustical highlight and seafloor reverberation areas, and based on this, it makes detection. The developed algorithm to achieve this consists of a migration and adaptation of a technique widely used in radar technology for detecting moving objects. This radar technique is known as the cell average-constant false alarm rate (CA-CFAR). This paper presents a drastic improvement of such approach by making an extension into 2-D analysis of the SSS image, in a way similar to integral image used in CA-CFAR detection for pulse Doppler radar. In this form, optimization of the computational effort is achieved. This new technique was called the accumulated cell average-constant false alarm rate in 2-D (ACA-CFAR 2-D). It was applied to pipeline detection and tracking with a very interesting degree of success. In addition, this technique provides similar results to image segmentation with respect to other frequently used approaches, but with much less computational resources and parameters to set. Its simplicity is a strong support of its robustness and accuracy. This feature makes it particularly attractive for using it in real-time applications, such as underwater robotics perception systems. This proposal was tested experimentally with acoustical data from SSS and the results detecting pipelines, and other shapes like sunken vessels or airplanes, are presented in this paper. Likewise, an experimental comparison with the results obtained with inverse undecimated discrete wavelet transform (UDWT) and active contours techniques is also presented.
27 December 2016 in Journal of Mathematical Imaging and Vision
Obstacle avoidance is one of the most relevant problems in mobile robotics field. The solution of this problem allows successful trajectory generation, detecting and avoiding those regions where a path is unable due to the presence of obstacles. The purpose of this paper is the application of the potential fields method for obstacles avoidance using a mechanical scanning sonar device for an autonomous underwater vehicle. Particular sonar's features are used to instantiate the proposed method. Different tests in simulated environment provided preliminary results that have proved its feasibility for the future implementation of this method in real environments, within the obstacle avoidance system of an autonomous underwater vehicle (AUV).
14 July 2017 in IEEE Journal of Oceanic Engineering
This paper describes a framework for object segmentation from sidescan sonar acoustic data. The current techniques consume a great deal of computational resources to accurately carry out object segmentation. They also involve the tuning of many parameters to obtain good quality images. This is due to the handling of the large data volume generated by these devices and environmental fluctuations such as salinity, density, temperature, and others variations. The framework proposed uses a migration and adaptation of a technique widely used in radar technology for detecting moving objects. This radar technique is known as order statistic-constant false alarm rate (OS-CFAR) applied in 2-D. OS-CFAR 2-D rank orders the samples obtained from a sliding window to make a segmentation of the image. This segmentation is done into several types of regions: acoustic highlight, shadow, and different seafloor reverberation areas. OS-CFAR 2-D is less sensitive than other methods to the presence of the speckle noise due to the use of order statistics. This proposal was contrasted experimentally on real images. Likewise, an experimental comparison with the results obtained with the undecimated discrete wavelet transform, active contours, Markov random field, and accumulated cell averaging CFAR applied in two dimensions technique is also presented.