Dog identification based on textural features and spatial relation of noseprint

•Development of a network model designed explicitly for dog noseprint detection and segmentation.•Successful identification and detection of crucial feature regions in dog noseprints.•Accurate segmentation of the scaly region in dog noseprints, allowing extraction of essential features from the scaly block.•Generation of an optimized parameter set for matching dog nose features, enhancing identification accuracy. This study proposes dog identification technology based on dog noseprints, which are equivalent to human fingerprints and possess unique characteristics. The aim is to utilize this technology for identifying and managing stray animals. The study presents three processing stages. In the first stage, YOLOv3 detects the dog's nose and nostril regions. The second stage involves enhancing the image's contrast and the contour of the scaly block using the multi-scale line detector. Finally, in the third stage, the shape and spatial features of the scaly block are extracted and utilized for dog identification. The study included a dataset of 276 dogs from multiple animal families and public shelters in Taiwan. The dataset was randomly divided into three groups to determine the optimal parameters for matching the dog noseprint via experimentation. Each dog identification group achieved an accuracy (ACC) exceeding 97.83 %, demonstrating that the proposed parameter-matching method offers high stability. Furthermore, in an additional experimental dataset consisting of dog noseprint images used for dog identification, the proposed method achieved an ACC exceeding 90.22 % in the Top 1 and 94.57 % in the Top 3. The ACC results across different groups consistently demonstrate the proposed method's ability to achieve high accuracy in dog identification. The source code and trained models are publicly available at: https://github.com/Chuen-HorngLin/Dog-Identification-Noseprint.