A Study on the Whole-skin Peeling System of a Snakehead Based on Three-dimensional Reconstruction

蛇头 人工智能 点云 卷积神经网络 计算机视觉 计算机科学 数学 算法 生物 渔业
作者
Ming Zhao,Min Lin,Ping Xu
出处
期刊:Applied Engineering in Agriculture [American Society of Agricultural and Biological Engineers]
卷期号:38 (5): 741-751
标识
DOI:10.13031/aea.15043
摘要

Highlights A 3D reconstruction algorithm is proposed to determine the snakehead area. The whole-skin peeling system of snakehead is designed by a convolutional neural network. An automatic whole-skin peeling device is presented with computer vision technology. Abstract. A whole-skin peeling system for a snakehead is designed in this article. The snakehead photos are taken from multiple angles and are first filtered to extract the image features, before reconstructing the sparse model. Then, the characteristic parameters of the snakehead are determined through the point cloud coordinates, and dense reconstruction is carried out based on sparse points. The accurate three-dimensional spatial coordinates of the snakehead are obtained by the octree algorithm and Poisson surface reconstruction algorithm. Based on the triangular meshing algorithm, the skin area of a snakehead is calculated to determine whether the snakehead meets the peeling conditions. A trained Faster Region Convolutional Neural Network (R-CNN) is used to identify the fin position on the snakehead reconstruction model and program the cutting route on the fish neck, while the cutting route of the fish belly is determined by selecting the area of the excretory hole through the ROI region of interest. Finally, the designed mechanical structure is used to peel the whole skin of the snakehead. The experimental results show that the maximum fluctuation rate of the snakehead area calculated by the proposed three-dimensional reconstruction algorithm is less than 4.0%, and the accuracy reaches 87.3%. The average precision (AP) for the determination method of the fish neck cutting position on the snakehead is 90.89%, which proves the effectiveness and feasibility of this method. The present results provide useful technical support for automating fish whole-skin peeling and shedding light on the accurate determination of the surface area and the extrication of the morphological feature parameters of other organisms. Keywords: Neural network, Point cloud, Snakehead, Skin peeling, Three-dimensional reconstruction.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
刚刚
蓝冰发布了新的文献求助50
2秒前
等待羿发布了新的文献求助10
2秒前
研友_VZG7GZ应助ZYP采纳,获得10
3秒前
木一发布了新的文献求助10
3秒前
4秒前
4秒前
4秒前
欢喜紫萍发布了新的文献求助10
4秒前
丘比特应助YJ采纳,获得10
5秒前
5秒前
顾顾关注了科研通微信公众号
5秒前
6秒前
6秒前
8秒前
地学韦丰吉司长完成签到,获得积分10
8秒前
9秒前
Rickpinkman发布了新的文献求助10
9秒前
英俊的铭应助碎碎采纳,获得10
9秒前
青烟完成签到 ,获得积分10
9秒前
10秒前
月亮很亮完成签到,获得积分10
11秒前
Rita发布了新的文献求助10
11秒前
mawanyu发布了新的文献求助10
12秒前
13秒前
13秒前
小羊羊发布了新的文献求助10
13秒前
13秒前
annie完成签到,获得积分10
13秒前
orixero应助狂野冰岚采纳,获得10
14秒前
CYY完成签到,获得积分20
14秒前
kun完成签到,获得积分10
14秒前
草莓味的星星完成签到,获得积分10
15秒前
万能图书馆应助月亮很亮采纳,获得10
15秒前
15秒前
15秒前
16秒前
16秒前
17秒前
怡然鸣凤发布了新的文献求助10
17秒前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
Modern Epidemiology, Fourth Edition 5000
Kinesiophobia : a new view of chronic pain behavior 5000
Molecular Biology of Cancer: Mechanisms, Targets, and Therapeutics 3000
Digital Twins of Advanced Materials Processing 2000
Weaponeering, Fourth Edition – Two Volume SET 2000
Signals, Systems, and Signal Processing 610
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 纳米技术 化学工程 生物化学 物理 计算机科学 内科学 复合材料 催化作用 物理化学 光电子学 电极 冶金 细胞生物学 基因
热门帖子
关注 科研通微信公众号,转发送积分 6018277
求助须知:如何正确求助?哪些是违规求助? 7606036
关于积分的说明 16158788
捐赠科研通 5165862
什么是DOI,文献DOI怎么找? 2765091
邀请新用户注册赠送积分活动 1746618
关于科研通互助平台的介绍 1635321