To search, Click below search items.


All Published Papers Search Service


Effect of local binary pattern on performance of iterative quantization in large-scale image retrieval


Mona Shakerdonyavi


Vol. 17  No. 3  pp. 150-161


Hashing is an efficient algorithm in order to approximate nearest neighbor in large-scale image datasets. Learning Binary Code is one of the key steps to improve the performance of these algorithms, however, it is still a challenge in this area. This article has improved hashing algorithm performance with the use of appropriate inputs. In the proposed method,image features extract by local binary pattern Then, thay use as input vector of iterativa quantization (ITQ) that leads to more compact codes, less memory and lower computational cost. The reasons behind these achivements are the binary nature and proper functioning of the local binary pattern. Finally in order to rerank results based on most similarity to query image, bit importance reranking method is used and then big weights are assigned to important bits and small weights are assigned to minor bits and weighted hamming distance is used to compute similarity between query image and top results.To evaluate performance of proposed method we use CIFAR-10 and MNIST as dataset and precision vs recall curve as evaluation criteria. The simulations compare the new algorithm with three state of the art and along the line algorithms from three points of view the hashing code size, memory space and computational cost, and the results demonstrate the effectiveness of the new approach.


content-based image retrieval, large scale, local binary pattern, iterative quantization, hashing algorithm