Learning the Relative Importance of Features in Image Data

Varde, Aparna S.; Rundensteiner, Elke A.; Javidi, Giti; Sheybani, Ehsan; Liang, Jianyii

doi:10.1109/icdew.2007.4400998

Cited by 18 publications

(11 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This study aims at reducing the dimensionality of dataset to reduce computational load in further processing [2]. The proposed method ranks features for learning a distance function in order to capture the semantics of the dataset [3]. It also uses the orthogonality properties of wavelets to decompose the dataset into spaces of coarse and detailed signals.…”

Section: Discussionmentioning

confidence: 99%

Digital Signal Processing for Analytics in Biostatistics & Biometric Applications

Sheybani

Javidi

2019

ABBA

Self Cite

View full text Add to dashboard Cite

All the biostatistics and biometric applications suffer from the effects of added noise due to their data dependency. The quality of data and impurities due to noise could affect the decisions made based on these datasets. Detecting anomalies caused by noisy datasets requires special preprocessing techniques that do not hurt the integrity of data. The authors have developed computationally low power, low bandwidth, and low-cost filters (DMAW) that will remove the noise, compress the dataset, and decompose the dataset so that a decision can be made by looking at different layers of data. This wavelet-based method is guaranteed to converge to a stationary point for both uncorrelated and correlated data. Presented here is the theoretical background with examples showing the performance and merits of this novel approach compared to other alternatives.

show abstract

Section: Discussionmentioning

confidence: 99%

Digital Signal Processing for Analytics in Biostatistics & Biometric Applications

Sheybani

Javidi

2019

ABBA

Self Cite

View full text Add to dashboard Cite

show abstract

“…Though domain experts may have subjective notions of similarity for comparison, they seldom have a distance function that ranks the image features based on their relative importance. The proposed method ranks features for learning such a distance function in order to capture the semantics of the images [3].…”

Section: Research and Methodologymentioning

confidence: 99%

“…The discrete Meyer adaptive wavelet (DMAW) is both translation-and scale-invariant and can represent a signal in a multi-scale format. While DMAW is not the best fit for entropy criterion, it is well suited for the proposed compression and cancellation purposes [8][9][10].…”

Section: Theory Of Dwt-based Filters For Noise Suppression and Order Reductionmentioning

confidence: 99%

Advanced Image Processing for Analytics in Biomedicine and Bioscience

Sheybani¹

2017

CTBEB

Self Cite

View full text Add to dashboard Cite

All medical imaging systems suffer from the effects of acquisition noise, channel noise, and fading. When decisions relevant to these image data are taken, any deviation from real values could affect the decisions made. Additionally, detecting anomalies from image data requires special processing to go behind the surface data. We have developed computationally low power, low bandwidth, and low cost filters (DMAW) that will remove the noise, compress the image, and decompose the image so that a decision can be made by looking at different layers of image data. This wavelet-based method is guaranteed to converge to a stationary point for both uncorrelated and correlated image data. Presented here is the theoretical background with examples showing the performance and merits of this novel approach compared to other alternatives.

show abstract

“…shape, color, texture or text annotations. If we are to use more than one type of feature, we have the problem of finding an appropriate weighting parameter w. Research on combining two or more features tends to either assume that labeled training data is available [21] [20], or it considers specialized domains where the value of w can be determined once and fixed forever. However it is clear that for the general problem of manuscript annotation the best value for w is highly data dependent.…”

Section: Background and Related Workmentioning

confidence: 99%

“…For the other extreme, imagine we are matching heraldic shields as in Figure 4.Right, here there is very little variation in shape (and none of it meaningful), and we would wish the algorithm to consider color only 1 . There are many existing techniques for learning this mixing parameter w, if we have access to subjective similarity judgments [20][16] [21]. While we would not rule out human interaction to refine a distance measure in an important domain, the scale of the problems we wish to eventually consider means that we would like to have a completely automated system to at least bootstrap the process and produce an initial high quality measure.…”

Section: Background and Related Workmentioning

confidence: 99%

Annotating Historical Archives of Images

Keogh

Shelton

et al. 2010

International Journal of Digital Library Systems

View full text Add to dashboard Cite

Recent initiatives like the Million Book Project and Google Print Library Project have already archived several million books in digital format, and within a few years a significant fraction of world' s books will be online. While the majority of the data will naturally be text, there will also be tens of millions of pages of images. Many of these images will defy automation annotation for the foreseeable future, but a considerable fraction of the images may be amiable to automatic annotation by algorithms that can link the historical image with a modern contemporary, with its attendant metatags. In order to perform this linking we must have a suitable distance measure which appropriately combines the relevant features of shape, color, texture and text. However the best combination of these features will vary from application to application and even from one manuscript to another. In this work we propose a simple technique to learn the distance measure by perturbing the training set in a principled way. We show the utility of our ideas on archives of manuscripts containing images from natural history and cultural artifacts.

show abstract

Learning the Relative Importance of Features in Image Data

Cited by 18 publications

References 10 publications

Digital Signal Processing for Analytics in Biostatistics & Biometric Applications

Digital Signal Processing for Analytics in Biostatistics & Biometric Applications

Advanced Image Processing for Analytics in Biomedicine and Bioscience

Annotating Historical Archives of Images

Contact Info

Product

Resources

About