Phytoplankton image texture classification based on dual-tree complex wavelet transform

作者：Yang Chenhui^*; Peng Mulin

来源：Journal of Computational Information Systems, 2012, 8(21): 8867-8875.

• Co-occurrence features
• Dual-tree complex wavelet transform
• Dual-tree complex wavelets
• High-pass
• Image texture classification
• Lazy learning
• Multi-scale features
• Multiscales
• Nearest Neighbor classifier
• Nearest neighbors
• Performance of classifier
• Statistical features
• Subbands
• Texture characteristics
• Texture classification
• Texture image

This paper proposes a supervised multiscale Nearest Neighbor texture classier for Phytoplankton Image. According to Dual-tree Complex Wavelet Transform (DT-CWT) characteristics and texture characteristics of circular algae, we have proposed a method to extract features based on dual-tree complex wavelet and principal component analysis (PCA). In this paper, we improve the performance of classifier by using both the magnitude and phase of the highpass sub bands of DT-CWT decomposition of a phytoplankton image. The statistical features and co-occurrence features of magnitude and phase are used to form a multiscale feature vector for each texture image. We use the Nearest Neighbor classifier which based on lazy learning strategy to recognize different species of circular algae. Experiment shows the validity of the proposed method.

• 出版日期2012
• 单位厦门大学