基于近红外深度图的游泳型鱼类摄食强度实时测量

针对水产养殖过程中精准投喂难题，提出一种基于近红外深度图的鱼类摄食活动强度评估方法。该方法不以目标跟踪为基础，不受养殖现场光照条件限制，无须依赖于清澈的水体与稳定背景，通过对深度数据直接处理，以最低的计算量实现鱼类前景目标提取及背景图像剔除，用新方法获取清晰的鱼类摄食图像。进而通过深度图的目标像素点总数判断抢食鱼类的数量，结合目标像素点的变化率，实时反映出鱼类摄食活跃程度；与传统的基于二维图像纹理特征的分析方法相比，该方法大幅度地降低了计算量，为养殖现场的实时测控提供了可实施方案。实验结果表明：近红外深度图不受养殖现场成像条件的限制，能以简洁的数据形式有效地表征鱼类的摄食规律，其方法对于分析在水面抢食浮性饲料，并在非摄食阶段栖息于水体底部的鱼类具有理想的分析效果。论文为鱼类行为分析提出一种新的技术手段，对精准投喂的应用具有积极的指导意义。

Evaluation of fish feeding intensity in aquaculture based on near-infrared depth image

Feeding is one of the primary factors determining efficiency and cost in aquaculture. In order to improve breeding efficiency, a new method of measuring fish feeding intensity based on depth image was proposed in this paper.This method is not limited by lighting conditions of aquaculture farms, and it does not rely on clear water or stable background. Through processing the depth data, the fish target foreground extraction and background image removal are realized with least amount of calculation, and clear fish feeding images are obtained by the depth image technology.The total number of the target pixels in the depth map is used to estimate the number of the feeding fish. Combined with the change rate of the target pixels, the feeding intensity of fish can be evaluated in real time. Compared with the traditional analysis based on two-dimensional image texture features, the method greatly reduces the amount of calculation task and provides an implementable scheme for the real-time measurement in aquaculture.The results show that the near-infrared depth image, which is not limited by the imaging conditions of the feeding site, can effectively represent the feeding behavior of fish.The method has ideal analysis effect for monitoring the fish that eat floating feed on the water surface and inhabit at the bottom of water body in non-feeding stage. This paper provides a new technical means for fish behavior analysis, which has positive guiding significance for the application of precision feeding.