Stochastic models for recognition of occluded targets

Bhanu, Bir; Lin, Yingqiang

doi:10.1016/s0031-3203(03)00182-1

Cited by 46 publications

(31 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Figure 1 shows the SAR images of the five targets used in the current exercise in two different poses. The MSTAR database is a standard dataset, which has been used in most of the reports found in the open literature, discussing the development of SAR ATR algorithms [8], [10]- [13].…”

Section: Database Usedmentioning

confidence: 99%

See 1 more Smart Citation

Validation of PCA and LDA for SAR ATR

Mishra

2008

TENCON 2008 - 2008 IEEE Region 10 Conference

View full text Add to dashboard Cite

Both principal component analysis (PCA) and linear discriminant analysis (LDA) have long been recognized as tools for feature extraction and data analysis. There has been reports in the open literature regarding the performance of both LDA and PCA as feature extractors in various types of classification and recognition problems. Many of the reports claim a better performance with LDA than with PCA. However, the grounds of comparison have mostly been quite narrow. In the current paper PCA and LDA based classifiers are evaluated for the problem of synthetic aperture radar based automatic target recognition problem. The results show that in terms of absolute performance, PCA outperforms LDA. Results of PCA based classifier are also found to be of higher confidence than those from LDA based classifiers, as observed from the error-bar analysis of the classifiers. With decreased amount of training dataset, the degradation in the performance of the classifiers are almost similar in nature. The current work concludes that LDA is not suitable for radar image based target recognition task. This is in line with reports from some works in the open literature which claim that the success of LDA will depend on the type of data and whether there is exhaustive data available during the training phase or not.

show abstract

Section: Database Usedmentioning

confidence: 99%

“…The reasons for choosing SAR ATR as the problem of choice are two fold. First of all SAR ATR has been one of the most challenging fields of pattern recognition for the past few decades [8]- [14]. Secondly, SAR ATR suffers from the problem of limited training dataset [14].…”

Section: Introductionmentioning

confidence: 99%

Validation of PCA and LDA for SAR ATR

Mishra

2008

TENCON 2008 - 2008 IEEE Region 10 Conference

View full text Add to dashboard Cite

show abstract

“…A great difficulty is that radar images are sensitive to target pose, so the scattering centers extracted from one image are only valid in its neighboring angular extent [21]. In [1], [12], [13], and [20], images at every 1 • angular interval are used to extract a scattering model, so the original data amount is still very large. In [19], a global scattering center model valid in a large angular extent is established by combining scattering centers extracted from 3-D images at every 3 • angular intervals, and the 3-D images are conveniently obtained by exploring the one-look ISAR ability of the shooting and bouncing ray technique.…”

mentioning

confidence: 99%

“…An unsolved problem is how to get scattering centers from a single SAR image, and any of such points would certainly not be the same as the 3-D ones. Projected 2-D scattering centers can be predicted from a 3-D scattering center model, but to match them with the scattering centers extracted from the test image is not an easy task [12], [13]. The main obstacles in this point-to-point matching include the following: 1) there are always redundant or missing points in both sets, so there may be no one-to-one correspondence between the two sets; 2) noise-caused error in the estimated scattering centers disturbs their match with the predicted scattering centers; and 3) pose estimation error changes the predicted scattering center feature, which further disturbs the match between the two sets.…”

mentioning

confidence: 99%

Automatic Target Recognition of SAR Images Based on Global Scattering Center Model

Zhou

Shi

Cheng

et al. 2011

IEEE Trans. Geosci. Remote Sensing

215

View full text Add to dashboard Cite

This paper proposes a synthetic aperture radar (SAR) automatic target recognition approach based on a global scattering center model. The scattering center model is established offline using range profiles at multiple viewing angles, so the original data amount is much less than that required for establishing SAR image templates. Scattering center features at different target poses can be conveniently predicted by this model. Moreover, the model can be modified to predict features for various target configurations. For the SAR image to be classified, regional features in different levels are extracted by thresholding and morphological operations. The regional features will be matched to the predicted scattering center features of different targets to arrive at a decision. This region-to-point matching is much easier to implement and is less sensitive to nonideal factors such as noise and pose estimation error than point-to-point matching. A matching scheme going through from coarse to fine regional features in the inner cycle and going through different pose hypotheses in the outer cycle is designed to improve the efficiency and robustness of the classifier. Experiments using both data predicted by a highfrequency electromagnetic (EM) code and data measured in the MSTAR program verify the validity of the method.Index Terms-Ground target, model based, scattering center, synthetic aperture radar (SAR) automatic target recognition (ATR).

show abstract

“…Bhanu and Lin [3] use a stochastic approach based on hidden Markov modelling for the recognition of occluded objects in synthetic aperture radar images and automatic target recognition.…”

Section: Literature Reviewmentioning

confidence: 99%

Restoration of Partially Occluded Shapes of Faces Using Neural Networks

Draganova

Lanitis

Christodoulou

Advances in Soft Computing

View full text Add to dashboard Cite

One of the major difficulties encountered in the development of face image processing algorithms, is the possible presence of occlusions that hide part of the face images to be processed. Typical examples of facial occlusions include sunglasses, beards, hats and scarves. In our work we address the problem of restoring the overall shape of faces given only the shape presentation of a small part of the face. In the experiments described in this paper the shape of a face is defined by a series of landmarks located on the face outline and on the outline of different facial features.We describe the use of a number of methods including a method that utilizes a Hopfield neural network, a method that uses Multi-Layer Perceptron (MLP) neural network, a novel technique which combines Hopfield and MLP together, and a method based on associative search. We analyze comparative experiments in order to assess the performance of the four methods mentioned above. According to the experimental results it is possible to recover with reasonable accuracy the overall shape of faces even in the case that a substantial part of the shape of a given face is not visible. The techniques presented could form the basis for developing face image processing systems capable of dealing with occluded faces.1

show abstract

Stochastic models for recognition of occluded targets

Cited by 46 publications

References 28 publications

Validation of PCA and LDA for SAR ATR

Validation of PCA and LDA for SAR ATR

Automatic Target Recognition of SAR Images Based on Global Scattering Center Model

Restoration of Partially Occluded Shapes of Faces Using Neural Networks

Contact Info

Product

Resources

About