Image classification is very important in pattern recognition and computer vision, where, for integrating final representation, feature pooling methods of the max-pooling, sum-pooling and average-pooling have been widely used. In this paper, we propose a new method called k-strongest responses (KSR) on the dictionary atoms for integrating the coding coefficients to generate the final representation, that, compared with the previous pooling methods, produces better performance for image classification task. Based on the KSR method, to improve classification accuracy and generate more compact and discriminative final representation, a new framework consisting of two-parts K-strongest response(KSR) and bag-of-features(BOF) is proposed. To evaluate the performance of the proposed method and framework, we apply it to Locality-constrained Linear Coding (LLC), Linear Distance Coding(LDC) and Sparse Coding(SC) by using two datasets from benchmarks of scene classification: 19-class satellite scene and UC Merced Land. The results show that the coding coefficients integrated by our method and framework are more discriminative than other methods.

Keywords