The invention discloses a meat duck
body size measuring
system based on multi-parameter collaborative acquisition. The
system comprises a meat duck multi-parameter acquisition module, a deep
reinforcement learning attitude
estimation module, a Fourier space
feature extraction module, a graph neural network relation reasoning module, a three-dimensional
point cloud analysis
processing module and a
body size parameter calculation output module. According to the
system, meat duck contour images, space coordinates and surface texture data are acquired through a multi-parameter acquisition module and transmitted to a posture
estimation module to recognize key posture feature points,
frequency domain feature vectors are extracted through a
feature extraction module, feature point association relationships are analyzed through a relationship reasoning module, a three-dimensional model is constructed according to the association relationships through a three-dimensional
point cloud analysis
processing module, and the meat duck contour images are obtained. And finally, the
body size parameter calculation and output module calculates and outputs parameters such as the oblique body length, the chest width, the chest depth, the shin length, the shin circumference and the
body height. According to the system, multi-parameter collaborative acquisition and multi-
algorithm fusion application are realized, the measurement efficiency and accuracy are improved, and meat duck body
size parameter measurement requirements in a large-scale breeding environment can be met.