Blue-green Algae Recognition Based on Rotation Invariant Uniform LBP and Color Information
Yingchun Zhang, Juanjuan Gu, Huimin Gao, Huabin Wang · 2020
Aiming at the problem of blue-green algae recognition in digital images, current algorithms ignore the difference between blue-green algae and water plants. In order to effectively distinguish between blue-green algae and other objects of the same color, this paper uses LBP histogram to extract the texture feature. But it takes a long time to recognize blue-green algae with the original LBP. Therefore, using the rotation invariant uniform LBP to extract image features. At this time, the number of feature vectors is reduced, significantly reduced recognition time. But it also led to the lack of some information, and accuracy of blue-green algae recognition is greatly reduced. Therefore, this paper concatenates the 64-dimension RGB histogram and the rotation invariant uniform LBP histogram to improve the recognition accuracy. The extracted features are then inputted into SVM with the INTER kernel function for final recognition. Experiments show that the proposed method is effective and efficient for blue-green algae recognition.