摘要
在家禽自动分割前对家禽胴体规格、质量进行分选。在悬挂输送线上设置展翅机、光影像扫描站、转载或卸载站及辅助设施。使用机器视觉识别技术,用高速视觉传感器对输送线上移动的禽胴体分14个分割部位进行快速扫描,采集时间最高可达0.1ms,对扫描数据快速计算分析,采用自适应禽胴体特征提取方法,对不同种类禽胴体检测,提取禽胴体的轮廓,密度,形态等体征,通过投影及衍射关系建立成像模型,确定高效率的图像处理加速算法,识别出不同级别的禽胴体范围。左右翅膀的面积差异大于30%,判定为翅折损。躯干近似轮廓的长轴方向倾斜角度大于10°,判定为挂单腿。数据转换成信号由分选系统接收后分别传送到指定的加工位置,最大处理能力达到了15 000只/h,平均准确率82%,误差不超过±30g。提高自动分割设备的定位准确性,提高自动分割禽胴体出成率,避免禽胴体局部损坏或缺失造成的禽胴体切割残次品现象,可以显著提高加工企业产品的附加值。
The poultry carcass specifications and quality were sorted before poultry automatic segmentation. The exhibition supercomputer, optical image scanning station, transshipment or unloading station and auxiliary facilities were set on the hanging transportation line. Machine vision recognition technology was used, and the high-speed visual sensor was adopted to quickly scan the 14 parts of the poultry carcass moving on the transportation line, and the acquisition time was up to 0.1 ms, and the scanning data was quickly calculated and analyzed. The adaptive poultry carcass feature extraction method was adopted, and different types of poultry carcass were detected, and the contours, density, morphology and other signs of the poultry carcass were extracted, and an imaging model was established by projection and diffraction relationships, and a highly efficient image processing acceleration algorithm was determined, and different grades of poultry carcass range were identified. The area difference between the left and right wings was greater than 30%, and it was judged as over broken. The inclination angle of the torso approximate contour in the long axis direction was greater than 10°, and it was judged as hanging single leg. The data was converted into signals and received by the sorting system and then transferred to the designated processing position. The maximum processing capacity was 15 000/h, and the average accuracy was 82%, and the error was no more than ±30 g. The positioning accuracy of automatic segmentation equipment was improved, and the rate of automatic segmentation of poultry carcasses was improved, and the phenomenon of poultry carcass cutting and defective products caused by partial damage or missing of carcass was averted, and it could significantly increase the added value of processing enterprise products.
作者
郭峰
刘立峰
张奎彪
刘辉
钟少辉
GUO Feng;LIU Lifeng;ZHANG Kuibiao;LIU Hui;ZHONG Shaohui
出处
《肉类工业》
2019年第11期31-40,共10页
Meat Industry
基金
国家重点研发计划项目(2018YFD0700803)
国家科技支撑计划项目(2015BAD19B06)
吉林省科技发展计划项目(20150309005GX)
关键词
家禽
胴体
分割
机器视觉
poultry
carcass
segmentation
machine vision
