System and method for estimating a length of underwater creatures

Patent metadata

Field	Value
Publication number	US-12524898-B2
Application number	US-202318190995-A
Country	US
Kind code	B2
Filing date	Mar 28, 2023
Priority date	Jul 22, 2022
Publication date	Jan 13, 2026
Grant date	Jan 13, 2026

Abstract

A method for estimating a length of underwater creatures and a system for estimating a length of underwater creatures are provided. The system includes a memory and a processor. The processor is electrically connected to the memory to load instructions in the memory to perform the method. The method includes: receiving a underwater image, in which the underwater image is captured by an image capturing device and includes a creature pattern of a target creature; performing an identification step on the creature pattern to obtain a creature lightness data set corresponding to the target creature; calculating a creature distance between the target creature and the image capturing device. In some embodiments, the identification step further obtains a creature structure data set of the target creature, and thus a creature length of the target creature is calculated in accordance with the creature distance and the creature structure data set.

First claim

What is claimed is:

1 . A method for estimating a length of underwater creatures, comprising:
performing a step for building a creature structural model, comprising:
providing a plurality of training images, wherein each of the training images comprises at least one training creature pattern and a plurality of training coordinate values corresponding to a plurality of creature feature points of the at least one training creature pattern; and
training a computer model to obtain the creature structural model by using the training images and the training coordinate values, wherein the computer model is a neural network model, a math model, or a statistical model; and
performing a step for on-line measuring, comprising:
receiving a plurality of underwater images of a breeding pond in a predetermined time period, wherein the underwater images are captured by using an image capturing device, and the underwater images comprise a plurality of creature patterns of a plurality of creatures;
performing identification on the creature patterns of the underwater images to obtain a plurality of creature structural data sets and a plurality of creature lightness data sets corresponding to the creatures by using the creature structural model;
determining a reference creature pattern from the creature patterns according to the creature structural data sets;
calculating a creature distance between each of the creatures and the image capturing device according to the creature lightness data set of each of the creature patterns and the creature lightness data set of the reference creature pattern; and
calculating a creature length of each of the creatures according to the creature distance between each of the creatures and the image capturing device and the creature structural data set of each of the creatures;
wherein the step for performing identification on the creature patterns of the underwater images by using the creature structural model comprises:
inputting each of the underwater images into the creature structural model to obtain a plurality of coordinate values of the creature feature points of each of the creature patterns;
wherein the step for calculating the creature distance between each of the creatures and the image capturing device according to the creature lightness data set of each of the creature patterns and the creature lightness data set of the reference creature pattern comprises:
calculating a lightness difference between the creature lightness data set of each of the creature patterns and the creature lightness data set of the reference creature pattern; and
calculating the creature distance between each of the creatures and the image capturing device according to the lightness difference of each of the creature patterns and a reference distance of the breeding pond.

2 . The method of claim 1 , wherein the step for calculating the creature length of each of the creatures according to the creature distance between each of the creatures and the image capturing device and the creature structural data set of each of the creatures comprises:
determining at least one part length of each of the creature patterns according to the coordinate values of the creature feature points of each of the creature patterns; calculating a creature observation length of each of the creatures according to the at least one part length of each of the creature patterns; calculating a length adjustment magnification according to the creature distance between each of the creatures and the image capturing device; and calculating the creature length of each of the creatures according to the creature observation length and the length adjustment magnification of each of the creatures.

3 . The method of claim 1 , wherein the creatures are shrimps, and the creature feature points comprise eyes of the shrimps, heads of the shrimps, viscera of the shrimps, bodies of the shrimps and tails of the shrimps.

4 . The method of claim 1 , wherein the training images and the underwater images are infrared images, and the image capturing device is an infrared camera.

5 . The method of claim 1 , wherein the predetermined time period is a day, and the underwater images are obtained by sampling an underwater video, and a sampling frequency of sampling the underwater video is one piece every three seconds.

6 . The method of claim 1 , further comprising:
calculating an average length of the creatures according to the creature lengths of the creatures; and storing the average length of the creatures in a database.

7 . The method of claim 1 , wherein the underwater images are captured by using one image capturing device only.

8 . A system for estimating a length of underwater creatures, comprising:
a memory configured to store a plurality of instructions; and a processor electrically connected to the memory to execute the instructions to: receive a plurality of underwater images of a breeding pond in a predetermined time period, wherein the underwater images are captured by using an image capturing device, and the underwater images comprise a plurality of creature patterns of a plurality of creatures; perform identification on the creature patterns of the underwater images to obtain a plurality of creature structural data sets and a plurality of creature lightness data sets corresponding to the creatures by using a creature structural model; determine a reference creature pattern from the creature patterns according to the creature structural data sets; calculate a creature distance between each of the creatures and the image capturing device according to the creature lightness data set of each of the creature patterns and the creature lightness data set of the reference creature pattern; and calculate a creature length of each of the creatures according to the creature distance between each of the creatures and the image capturing device and the creature structural data set of each of the creatures; wherein when the processor performs identification on the creature patterns of the underwater images by using the creature structural model, the processor is configured to:
input each of the underwater images into the creature structural model to obtain a plurality of coordinate values of a plurality of creature feature points of each of the creature patterns;
wherein when the processor calculates the creature distance between each of the creatures and the image capturing device according to the creature lightness data set of each of the creature patterns and the creature lightness data set of the reference creature pattern, the processor is configured to:
calculate a lightness difference between the creature lightness data set of each of the creature patterns and the creature lightness data set of the reference creature pattern; and
calculate the creature distance between each of the creatures and the image capturing device according to the lightness difference of each of the creature patterns and a reference distance of the breeding pond;
wherein the predetermined time period is a day, and the underwater images are obtained by sampling an underwater video, and a sampling frequency of sampling the underwater video is one piece every three seconds.

9 . The system of claim 8 , wherein when the processor calculates the creature length of each of the creatures according to the creature distance between each of the creatures and the image capturing device and the creature structural data set of each of the creatures, the processor is configured to:
determine at least one part length of each of the creature patterns according to the coordinate values of the creature feature points of each of the creature patterns; calculate a creatu

Assignees

• National Sun Yat Sen University

Inventors

• Huang Ing-Jer
• Chang Yun-Nan

Classifications

• G06V20/05 Primary

  Underwater scenes · CPC title

• G06T2207/10048

  Infrared image · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title