Partial-Aliasing Correlation Filters

Fernández, Julio César Ramos; Kumar, B. V. K. Vijaya

doi:10.1109/tsp.2014.2385667

Cited by 5 publications

(4 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, aliasing effect introduced due to circular correlation instead of linear correlation was identified by the researchers as a root cause of performance issues in correlation filters. Partial-aliasing correlation filters were introduced to optimise the performance by producing sharper correlation peaks [30]. Efficacy of CPR has also been exploited for human action recognition in some recent studies [31][32][33][34][35][36], where these filters have shown quite promising results.…”

Section: Latest Trends In Cpr Filtersmentioning

confidence: 99%

“…The lexicographical scanning method was adopted while testing the filter response with a step size of three pixels (in order to have an in-depth view of the proposed filter response). DoG enhanced, logarithmic map of each sub-image extracted out of test image was correlated with the reference image using (30). The output correlation plane corresponding to the sub-image producing max r i was earmarked as the strongest output depicting the occurrence of the target in that specific area.…”

Section: Performance Evaluation Frameworkmentioning

confidence: 99%

See 1 more Smart Citation

Composite filtering strategy for improving distortion invariance in object recognition

Awan

Rehman

Bakhshi

2018

IET Image Processing

View full text Add to dashboard Cite

Section: Latest Trends In Cpr Filtersmentioning

confidence: 99%

Section: Performance Evaluation Frameworkmentioning

confidence: 99%

Composite filtering strategy for improving distortion invariance in object recognition

Awan

Rehman

Bakhshi

2018

IET Image Processing

View full text Add to dashboard Cite

“…The CPR localizes the target when the SVM allowed generalization in the input. Further enhancement in the CPR performance is introduced through partial-aliasing correlation filters [25]. These filtering techniques ensure sharper peaks in the presence of a target/object.…”

mentioning

confidence: 99%

Weight Quantization Retraining for Sparse and Compressed Spatial Domain Correlation Filters

et al. 2021

View full text Add to dashboard Cite

Using Spatial Domain Correlation Pattern Recognition (CPR) in Internet-of-Things (IoT)-based applications often faces constraints, like inadequate computational resources and limited memory. To reduce the computation workload of inference due to large spatial-domain CPR filters and convert filter weights into hardware-friendly data-types, this paper introduces the power-of-two (Po2) and dynamic-fixed-point (DFP) quantization techniques for weight compression and the sparsity induction in filters. Weight quantization re-training (WQR), the log-polar, and the inverse log-polar geometric transformations are introduced to reduce quantization error. WQR is a method of retraining the CPR filter, which is presented to recover the accuracy loss. It forces the given quantization scheme by adding the quantization error in the training sample and then re-quantizes the filter to the desired quantization levels which reduce quantization noise. Further, Particle Swarm Optimization (PSO) is used to fine-tune parameters during WQR. Both geometric transforms are applied as pre-processing steps. The Po2 quantization scheme showed better performance close to the performance of full precision, while the DFP quantization showed further closeness to the Receiver Operator Characteristic of full precision for the same bit-length. Overall, spatial-trained filters showed a better compression ratio for Po2 quantization after retraining of the CPR filter. The direct quantization approach achieved a compression ratio of 8 at 4.37× speedup with no accuracy degradation. In contrast, quantization with a log-polar transform is accomplished at a compression ratio of 4 at 1.12× speedup, but, in this case, 16% accuracy of degradation is noticed. Inverse log-polar transform showed a compression ratio of 16 at 8.90× speedup and 6% accuracy degradation. All the mentioned accuracies are reported for a common database.

show abstract

“…Although, a remarkable class segregation accuracy is achieved by the enhanced versions of correlation filters negative impact of clutters on the correlation outputs still remains the problem not handled in CPR paradigm [5][6][7][8][9]. Some of the known impacts caused by the presence of clutters include pronounced side-lobe generation and reduction in height of correlation peak.…”

mentioning

confidence: 99%

Active contour‐based clutter defiance scheme for correlation filters

et al. 2019

View full text Add to dashboard Cite

The negative impact of clutters on correlation output still remains the problem not handled in correlation pattern recognition (CPR) paradigm. Some of the known impacts caused by the presence of clutters include pronounced side-lobe generation and reduction in height of correlation peak. The occurrence of the side-lobes results in the distortion of actual correlation output while making the class decision a difficult task. Whereas, reduction in correlation height may cause class decisions to go wrong. The authors propose a novel clutter defiance strategy based on object segmentation through active contours approach thus evading a negative effect on the correlation output. Instead of allowing clutters to participate in CPR-based classification decision they recommend blocking irrelevant details to avoid distortion in the correlation process. An appropriate tweaking in the optimal logarithmic maximum average correlation height filter design yields a performance gain up to 80% in the correlation peak to side-lobe ratio, thus making it well pronounced to classify the true class objects correctly. The improvement remains consistent throughout the range of in-plane angular distortions against synthetically cluttered challenging images.

show abstract

Partial-Aliasing Correlation Filters

Cited by 5 publications

References 29 publications

Composite filtering strategy for improving distortion invariance in object recognition

Composite filtering strategy for improving distortion invariance in object recognition

Weight Quantization Retraining for Sparse and Compressed Spatial Domain Correlation Filters

Active contour‐based clutter defiance scheme for correlation filters

Contact Info

Product

Resources

About