Abstract: Abstract: Monitoring behavior of pigs in a pen is possible both in group and at individual level. Data analysis at individual level, however, has more advantages. Identification of pigs is a necessary step towards analyzing the different behaviors of pigs individually. Some current computer vision systems that are used for video surveillance of group-housed pigs require that the pigs be marked. In this paper, using a top-view video sequence of group-housed pigs, a machine-vision technology method for recognizing individual pig is proposed. First, to recognize each individual pig, foreground detection and target extraction are conducted on a top-view video sequence of the group-housed pigs. Second, the training samples are established, and the color, texture and shape of the individual pig are extracted; through the combination of these features, a feature vector representing an individual pig is then built. Third, the combined features are fused using the Isomap algorithm, which reduces the feature dimension on the basis of the maximum retention of the effective recognition information. Finally, the features are trained and recognized using a support vector machine (SVM) classifier with an optimal kernel function. The videos used in the present study are collected from pig farms of the Danyang Rongxin Nongmu Development Company, which is the experimental base for the key discipline of Jiangsu University, i.e. agricultural electrification and automation. The pigs are monitored in a reconstructed experimental pigsty. The pigsty is 1 m high, 3.5 m long and 3 m wide. A camera is located above the pigsty with the height of 3 m over the ground. The camera is the FL3-U3-88S2C-C with an image resolution of 1760 × 1840 pixels from the Grey Point Company. The videos are captured from 8 AM to 5 PM. Over 5 days randomly chosen, we collect 6 sections of videos every day at random time, so there are a total of 30 videos randomly chosen in audio video interleaved (AVI) format. The frame frequency of each video is 25 fps with the duration of approximately 120 s. Among the 90 000 frames (5 days × 6 videos × 120 s × 25 fps), 900 frames satisfying the requirement of experimental conditions are selected. The software MATLAB 2012b is adopted. The experimental results show that the proposed method is effective and the highest recognition rate of pigs is 92.88%. In this paper, a method for recognizing group-housed pigs individually from a top-view video sequence is explored based on the machine vision, which differs from traditional radio frequency identification (RFID) of individual pig. This study provides a new idea for the recognition of individual pig without stressing the animals, and lays a foundation for further analysis of the behavior of individual pig.