Robust score-based feature vectors for algorithm fusion

Dobeck, Gerald J.

doi:10.1117/12.542731

Cited by 5 publications

(2 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fourteen difficult images, including the two shown in the Figures 3 and 4, were selected for analysis. These were normalized by the two normalizers and processed by the same D/C algorithm [2][3][4][5] . Under both normalizers, the D/C algorithm detected and classified the same mines; however there was marked reduction in the false alarms with the SFBF normalized image as shown in the Table 1.…”

Section: Impact On False Alarm Reductionmentioning

confidence: 99%

Image normalization using the serpentine forward-backward filter: application to high-resolution sonar imagery and its impact on mine detection and classification

Dobeck

2005

SPIE Proceedings

Self Cite

View full text Add to dashboard Cite

In high-resolution sonar imagery, large variations in image background can make it difficult to reliably find targets. These variations result from irregular illumination of the sea floor, which is caused by sonar platform motion. Contributing to this problem is the fact that the spatial frequencies of the varying background can be similar to those of the target. Consequently, image-processing methods that attempt to segment image regions associated with target highlight (or shadow), are often fooled by bright (or dark) target-size patches of background. This typically results in an increase of the number of false alarms. This paper describes two image normalization methods: the Cross-Range Forward-Backward filter and the Serpentine Forward-Backward filter. Results are presented that show the impact of the image normalization on reducing false alarms.

show abstract

Section: Impact On False Alarm Reductionmentioning

confidence: 99%

Image normalization using the serpentine forward-backward filter: application to high-resolution sonar imagery and its impact on mine detection and classification

Dobeck

2005

SPIE Proceedings

Self Cite

View full text Add to dashboard Cite

show abstract

“…For example, (1) majority voting (or more generally m out of n) where the detections can be conditioned on thresholds applied to the scores, (2) computing the sum of the algorithm scores and comparing the sum to a threshold, or (3) computing a linear combination of the scores and comparing the weighted sum to a threshold. And there are many others [6]- [12].…”

Section: Score-based Fusion Rulementioning

confidence: 99%

A Probabilistic Model for Score-based Algorithm Fusion

Dobeck

Proceedings of OCEANS 2005 MTS/IEEE

View full text Add to dashboard Cite

─ A probabilistic model is described that helps explain how Score-based Algorithm Fusion achieves significant false alarm reduction. The two-class detection and classification (DC) problem is considered: the target and the non-target. Multiple DC algorithms, based on fundamentally different DC methodologies, process the same sensor data looking for targetlike objects. Each object detected and classified as target-like by a given algorithm is assigned a positive score, which indicates the degree to which the algorithm considers the object target-like. Score-based Algorithm Fusion is the fusion of multiple detection & classification algorithms where only the scores of the individual algorithms are used to make a final determination on whether an object is a target or not. Despite the fact that only the scores are used in the fusion process, false alarm reduction has been remarkable while still preserving a high probability of target detection and classification. A probabilistic model is presented in this paper that supports this observation.

show abstract

The K-Nearest Neighbor Attractor-based Neural Network and the Optimal Linear Discriminatory Filter Classifier

Dobeck

2006

Oceans 2006

View full text Add to dashboard Cite

Robust score-based feature vectors for algorithm fusion

Cited by 5 publications

References 18 publications

Image normalization using the serpentine forward-backward filter: application to high-resolution sonar imagery and its impact on mine detection and classification

Image normalization using the serpentine forward-backward filter: application to high-resolution sonar imagery and its impact on mine detection and classification

A Probabilistic Model for Score-based Algorithm Fusion

The K-Nearest Neighbor Attractor-based Neural Network and the Optimal Linear Discriminatory Filter Classifier

Contact Info

Product

Resources

About