基于改进Mask RCNN和SVR的无接触梭子蟹体质量预测研究
PDF下载 (296)唐 潮,胡海刚,史 策,钱云霞.基于改进Mask RCNN和SVR的无接触梭子蟹体质量预测研究[J].宁波大学学报(理工版),2021,34(2):31-41.DOI:
TANG Chao,HU Haigang,SHI Ce,QIAN Yunxia.Weight estimation of non-contact swimming crabs based on improved Mask RCNN and SVR[J].Journal of Ningbo University(Natural Science & Engineering Edition),2021,34(2):31-41.DOI:
| Title: | Weight estimation of non-contact swimming crabs based on improved Mask RCNN and SVR |
| 作者: | 唐 潮, 胡海刚, 史 策, 钱云霞 |
| Author(s): | TANG Chao, HU Haigang, SHI Ce, QIAN Yunxia |
| 关键词: | 梭子蟹; 视觉特征; 体质量; 无接触测量 |
| Keywords: | Mask RCNN; swimming crab; visual feature; weight; non-contact measurement |
| 分类号: | TP274 |
| 文献标识码: | A |
| 摘要: | 提出了一个改进Mask RCNN目标检测算法用以对养殖梭子蟹进行视觉特征测量. 通过在养殖区域采集梭子蟹图像, 用上位机识别梭子蟹旋转角度以及甲长和甲宽方向, 对输出的Mask进行模板修补, 提高模板内区域的置信度. 通过图像-实景对应关系换算梭子蟹的真实尺寸, 并估算其投影面积、甲宽与甲长, 结果准确率高于85%. 同时, 对视觉算法得到的梭子蟹尺寸特征与其体质量进行拟合, 引入k-means聚类, 实现双模型支持向量回归机(SVR)预测结构. 通过差分进化算法对SVR适应度函数进行寻优, 设计了随迭代次数、寻优效果同步变化的缩放因子, 以及适者更易生存策略的交叉概率因子, 以验证改进算法的寻优能力. 测试时, 对新传入的数据首先进行归一化处理, 然后判断所归属的聚类中心, 再传至相应的SVR模型进行预测. 测试结果相对误差小于18%. |
| Abstract: | The data of farmed crabs mainly rely on manual measurement, which is both labor intensive and resource wasting. In this paper, an improved Mask RCNN size measurement algorithm is proposed, and the original Mask based on probability output from Mask RCNN is modified according to the template obtained in advance to achieve more accurate size estimation. First, crab images are captured by using cameras, which are then segmented with the Mask RCNN algorithm. After applying a second-order moment algorithm on the mask, the template is transferred and rotated into the center of the original mask, and the corresponding part is modified by adding and subtracted benefits, so as to obtain a more accurate image of the carapace. The real size of the crab is converted through the image-real scene mapping, and the area, width and length of the crab carapace are estimated with an accuracy of more than 85%. The size obtained by the visual algorithm is fitted to the weight of the swimming crab. In order to improve the matching degree of the model, the k-means clustering is introduced to implement a dual-model support vector regression machine (SVR) prediction structure, which uses the differential evolution algorithm to optimize the fitness function of the SVR. In the test process, after normalizing the newly input data, the cluster center to which it belongs is first determined, and then the corresponding SVR model is applied for prediction. The test set results show that the relative error remains below 18%. |
| 参考文献 /References: | [1] 曹广斌, 蒋树义, 韩世成, 等. 工厂化水产养殖中的自动控制技术[J]. 水产学杂志, 2011, 24(1):60-64. [2] 周显青, 孙儒泳, 牛翠娟. 应激对水生动物生长、行为和生理活动的影响[J]. 动物学研究, 2001, 22(2):154- 158. [3] Mathiassen J R, Misimi E, Bond? M, et al. Trends in application of imaging technologies to inspection of fish and fish products[J]. Trends in Food Science & Technology, 2011, 22(6):257-275. [4] 段延娥, 李道亮, 李振波, 等. 基于计算机视觉的水产动物视觉特征测量研究综述[J]. 农业工程学报, 2015, 31(15):1-11. [5] Zion B. The use of computer vision technologies in aquaculture – A review[J]. Computers and Electronics in Agriculture, 2012, 88:125-132. [6] 周莉莉, 姜枫. 图像分割方法综述研究[J]. 计算机应用研究, 2017, 34(7):1921-1928. [7] Viazzi S, van Hoestenberghe S, Goddeeris B M, et al. Automatic mass estimation of Jade perch Scortum barcoo by computer vision[J]. Aquacultural Engineering, 2015, 64:42-48. [8] 王文静, 徐建瑜, 吕志敏, 等. 基于机器视觉的水下鲆鲽鱼类质量估计[J]. 农业工程学报, 2012, 28(16):153- 157. [9] 张超, 徐建瑜, 王文静. 基于机器视觉的梭子蟹质量估计方法研究[J]. 宁波大学学报(理工版), 2014, 27(2):49- 51. [10] 姚霆, 张炜, 刘金根. 基于深度学习的图像分割技术[J]. 人工智能, 2019, 6(2):66-75. [11] 骆桂兰, 陈军, 王会聪, 等. 水产养殖过程中计算机视觉的应用[J]. 畜牧与饲料科学, 2017, 38(12):91-92. [12] 房永峰. 基于深度学习的牲畜目标检测与跟踪算法研究[D]. 合肥: 中国科学技术大学, 2019. [13] Redmon J, Divvala S, Girshick R, et al. You only look once: Unified, real-time object detection[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016:779-788. [14] 赵德安, 刘晓洋, 孙月平, 等. 基于机器视觉的水下河蟹识别方法[J]. 农业机械学报, 2019, 50(3):151-158. [15] Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014:580-587. [16] He K M, Sun J, Tang X O. Single image haze removal using dark channel prior[C]//2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009:20-25. [17] 王静安, 陈开健, 刘小燕, 等. 大通湖水域河蟹体长、体重及头胸甲关系研究[J]. 湖南文理学院学报(自然科学版), 2017, 29(1):35-39. [18] 张超. 计算机视觉技术在梭子蟹养殖中的应用研究[D]. 宁波: 宁波大学, 2014. [19] 唐杨捷, 胡海刚, 张刚, 等. 基于计算机视觉和GA- SVM的梭子蟹体重预测[J]. 宁波大学学报(理工版), 2019, 32(1):32-37. [20] 陈军. 基于机器视觉的汽车油泵支撑杆尺寸实时检测系统[D]. 杭州: 浙江大学, 2017. [21] Ren S Q, He K M, Girshick R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149. [22] 高保全, 刘萍, 李健, 等. 三疣梭子蟹形态性状对体重影响的分析[J]. 海洋水产研究, 2008, 29(1):44-50. [23] 高天龙, 王燕飞, 鲍雪宁, 等. 单体筐养三疣梭子蟹的蜕壳与生长规律研究[J]. 生物学杂志, 2016, 33(3):41- 46. [24] 高岳林, 刘军民. 差分进化算法的参数研究[J]. 黑龙江大学自然科学学报, 2009, 26(1):81-85. [25] 颜学峰, 余娟, 钱锋, 等. 基于改进差分进化算法的超临界水氧化动力学参数估计[J]. 华东理工大学学报(自然科学版), 2006, 32(1):94-97; 124. [26] 纪震, 廖惠连, 吴青华. 粒子群算法及应用[M]. 北京: 科学出版社, 2009:74-78. |
| 备注/Memo: | 收稿日期: 2020－05－18. 宁波大学学报（理工版）网址: http://journallg.nbu.edu.cn/ 基金项目: 浙江省公益技术项目（2017C32014）; 浙江省重点研发计划（2019C02055）; 宁波市科技富民项目（2017C10006）; 宁波市自然科学基金（2019A610424）. 宁波大学学报（理工版）网址：http://journallg.nbu.edu.cn/第一作者: 唐潮（1995－）, 男, 江苏宿迁人, 在读硕士研究生, 主要研究方向: 计算机视觉. E-mail: 2906941684@qq.com *通信作者: 胡海刚（1966－）, 男, 浙江舟山人, 高级实验师, 主要研究方向: 船舶工程安全技术与机电控制. E-mail: huhaigang@nbu.edu.cn |
