Abstract:In order to obtain the three-dimensional (3D) distribution of fish in a marine ranching, a method using imaging sonar to calculate the 3D coordinates of fish is proposed in this paper. The imaging sonar used in this re-search is a Dual-frequency Identification sonar (DIDSON). It is a multi-beam sonar that uses acoustic lens to form individual beams. It constructs a high-definition two-dimensional image for target detection by transmitting ultra-sonic beams underwater and receiving the echo signals. It sets a new standard for excellence in underwater vision in black and turbid waters due to obtaining near-video quality dynamic images for the identification of objects un-derwater. In addition, split-beam echo-sounders have also been used on some occasions to monitor the movements of fishes. Despite the improvements achieved in fish monitoring techniques, the interpretation and classification of the data collected by the traditional acoustic techniques are often challenging and require extensive experience and effort. The DIDSON bridges the gap between existing fisheries-assessment sonar and optical systems. In a DID-SON, 96 transducer elements constitute a linear array and each element both transmits and receives acoustic beams such that echo amplitude is determined by the intensity of the reflected signal. It can obtain the distance and azi-muth of the target from sonar images, but it is unable to acquire the elevation of the target from the images. To overcome this difficulty and obtain the fish distribution, a new method is proposed. Firstly, the sonar is fixed on the outside of the ship′s rail and submerged in the water to collect fish′s information through the investigation on navigation. At the same time, the beam emission direction is on the same plane with the sonar′s moving direction. After data collection, image processing is conducted, including image construction, background elimination and target extraction from horizontal field-of-view. Target association based on Interacting Multiple Model Joint Prob-abilistic Data Association Filtering (IMMJPDAF) is carried out to deal with the extracted targets, thus the rela-tions of one target in different frame images can be obtained. The 3D target coordinates are acquired according to the spatial geometric relationship between the positions in two consecutive frame images. Finally, multiple target trajectories in 3D space and the depth distribution of targets are obtained through the correlation algorithm. The ex-periment was carried out in Dishui Lake which is located in Shanghai. Experimental results showed that the pro-posed method can effectively acquire the fish movement tracks in 3D space underwater and the distributions in depth. It also showed that most fish swam in the depth 3-5 meters. It will help to analyze the fish behavior and provide technical support for fishery resource assessment in a marine ranching.

Research on A Binocular Fish Dimension Measurement Method Based on Instance Segmentation and Fish Tracking

Stereo Vison and Mask-RCNN Segmentation Based 3D Points Cloud Matching for Fish Dimension Measurement

Fish Keypoints Detection for Ecology Monitoring Based on Underwater Visual Intelligence

Multi-Scale Fish Segmentation Refinement and Missing Shape Recovery

Coarse-To-Fine Segmentation Refinement and Missing Shape Recovery for Halibut Fish

A Robust Underwater Multiclass Fish-School Tracking Algorithm

Feature Fusion of Body Surface and Motion-Based Instance Segmentation for High-Density Fish in Industrial Aquaculture

Online Fish Tracking with Portable Smart Device for Ocean Observatory Network

An Automatic Detection and Statistical Method for Underwater Fish Based on Foreground Region Convolution Network (FR-CNN)

Three-dimensional Distribution of Fish Using an Imaging Sonar

Closed-Loop Tracking-by-Detection for ROV-Based Multiple Fish Tracking

Multi-class Fish Stock Statistics Technology Based on Object Classification and Tracking Algorithm

Accurate Fish Detection under Marine Background Noise Based on the Retinex Enhancement Algorithm and CNN

TOF-assisted binocular vision accuracy improvement method for underwater fish size inspection

Key point detection method for fish size measurement based on deep learning

Segmentation and measurement scheme for fish morphological features based on Mask R-CNN

In-Water Fish Body-Length Measurement System Based on Stereo Vision

Fish Trajectory Extraction Based on Object Detection

A Fish Detection and Tracking Method Based on Improved Interframe Difference and YOLO-CTS

Feature Point Based 3D Tracking of Multiple Fish from Multi-View Images

Fish Tracking and Segmentation From Stereo Videos on the Wild Sea Surface for Electronic Monitoring of Rail Fishing