Adaptive quasiconformal kernel metric for image retrieval

Heisterkamp, Douglas R.; Peng, Jing; Dai, Honghua

doi:10.1109/cvpr.2001.990987

Cited by 22 publications

(17 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As discussed above, relevance feedback is commonly used in CBIR systems for improving the retrieval performance [10,7,15,9,6,5,16,17,14].…”

Section: Discussionmentioning

confidence: 99%

“…The same approach has also been applied to a correlation-based metric [7,8], which usually outperforms Euclidean-based measures. In [9], the authors presented an approach to generate an adaptive quasiconformal kernel distance metric based on relevance feedback. Dong and Bhanu [10] proposed a new semi-supervised expectation-maximization (EM) algorithm for image retrieval tasks, with the image distribution in the feature space modeled as Gaussian mixtures.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Kernel-based distance metric learning for content-based image retrieval

Chang

Yeung

2007

Image and Vision Computing

View full text Add to dashboard Cite

For a specific set of features chosen for representing images, the performance of a content-based image retrieval (CBIR) system depends critically on the similarity or dissimilarity measure used. Instead of manually choosing a distance function in advance, a more promising approach is to learn a good distance function from data automatically. In this paper, we propose a kernel approach to improve the retrieval performance of CBIR systems by learning a distance metric based on pairwise constraints between images as supervisory information. Unlike most existing metric learning methods which learn a Mahalanobis metric corresponding to performing linear transformation in the original image space, we define the transformation in * Corresponding author: Dr. Dit-Yan Yeung, dyyeung@cs.ust.hk, +852-2358-1477 1 This is the Pre-Published Version the kernel-induced feature space which is nonlinearly related to the image space.Experiments performed on two real-world image databases show that our method not only improves the retrieval performance of Euclidean distance without distance learning, but it also outperforms other distance learning methods significantly due to its higher flexibility in metric learning.

show abstract

“…As discussed above, relevance feedback is commonly used in CBIR systems for improving the retrieval performance [10,7,15,9,6,5,16,17,14].…”

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Kernel-based distance metric learning for content-based image retrieval

Chang

Yeung

2007

Image and Vision Computing

View full text Add to dashboard Cite

show abstract

“…A variety of retrieval methods have been developed that exploit relevance feedback to create flexible retrieval metrics, thereby improving accuracy. Kernel methods for image retrieval with relevance feedback have been presented and shown to outperform weighted Euclidean distances [3,9,29]. The relevance feedback methods help adapt the distance function to the user's sense of relevance.…”

Section: Introductionmentioning

confidence: 99%

“…While approximation based index techniques [25,27] have shown promise in large databases with high dimensions and support relevance feedback retrieval with linear weightings (i.e., database objects still maintain the relative positions along each axis) [10,15,17], they are unable to support nonlinear transformations, such as kernel distances [9,3,29]. A vector approximation file (VA-File) takes a signature or filter approach to indexing data [25].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Kernel Vector Approximation Files for Relevance Feedback Retrieval in Large Image Databases

Heisterkamp

Peng

2005

Multimed Tools Appl

View full text Add to dashboard Cite

Many data partitioning index methods perform poorly in high dimensional space and do not support relevance feedback retrieval. The vector approximation file (VA-File) approach overcomes some of the difficulties of high dimensional vector spaces, but cannot be applied to relevance feedback retrieval using kernel distances in the data measurement space. This paper introduces a novel KVA-File (kernel VA-File) that extends VA-File to kernel-based retrieval methods. An efficient approach to approximating vectors in an induced feature space is presented with the corresponding upper and lower distance bounds. Thus an effective indexing method is provided for kernel-based relevance feedback image retrieval methods. Experimental results using large image data sets (approximately 100,000 images with 463 dimensions of measurement) validate the efficacy of our method.

show abstract