Composite filtering strategy for improving distortion invariance in object recognition

Awan, Ahmed Bilal; Rehman, Saad; Bakhshi, Asim Dilawer

doi:10.1049/iet-ipr.2017.1147

Cited by 7 publications

(5 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…EEMACH [20] and its derivatives [44] showed a remarkable performance as compared to other CPR filters. Literature shows their superior clutter-rejection capability as compared to other methods, and experiments, which have shown better results, are conducted taking N v = 1 and not at other value of N v .…”

Section: Cpr Filter Implementations and Settingmentioning

confidence: 99%

Weight Quantization Retraining for Sparse and Compressed Spatial Domain Correlation Filters

et al. 2021

Self Cite

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

Section: Cpr Filter Implementations and Settingmentioning

confidence: 99%

Weight Quantization Retraining for Sparse and Compressed Spatial Domain Correlation Filters

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…Recently Awan et al proposed a combined framework of the EEMACH filter and the DOG filter to develop a system which can give a pronounced peak in the presence of background clutter and distortions [7]. The proposed filter gave enhanced target recognition results leading to a higher percentage of correct automated decisions in all situations as compared to previously proposed correlation filters.…”

Section: Background Reviewmentioning

confidence: 99%

Hardware design of correlation filters for target detection

Akbar¹,

Tehsin²,

Rehman³

et al. 2019

Pattern Recognition and Tracking XXX

Self Cite

View full text Add to dashboard Cite

Correlation filters have been implemented in software and have proven very effective for automatic target detection, biometric verification and security applications. In this paper, these filters are implemented in hardware keeping in view their impotance in real time applications. Hardware implementation results are placed in comparison with results generated through software. These results vary by as little as 10-4 which is demonstrated in the experimental section. The hardware design of these filters is implemented in LabView which can be subsequently employed in real-time security applications. This design may be expanded for other advanced variants of correlation filters in future work.

show abstract

“…The matched filter is arguably the most ancient and straightforward filter [5]. A wide variety of filter design approaches have been put forth in the literature to expand this fundamental concept, such as synthetic discriminant function (SDF) filter [6], phase-only filter (POF) [7][8][9][10][11][12][13][14][15][16][17], minimum average correlation energy (MACE) filter [18][19][20], maximum average correlation energy (MACH) filter [21][22][23][24][25][26][27], extended MACH filter [28][29][30][31][32][33][34], wavelet-modified MACH filter [35], maximum margin correlation filter [36], neural network-based correlation filter [37,38], and segmented composite filter [39][40][41].…”

Section: Introductionmentioning

confidence: 99%

Wavelet-modified segmented phase-only filter for in-plane rotated object recognition

Kumar,

Nishchal,

Alfalou

2024

Phys. Scr.

View full text Add to dashboard Cite

In this paper, we propose and implement a wavelet-modified segmented phase-only filter for in-plane rotated object recognition. The wavelet filter is combined with the segmented phase-only filter to significantly enhance the correlation capability of the optical processor. To verify the consistency of the results, we have calculated the performance measure parameters such as peak-to-correlation energy, peak-to-sidelobe ratio, discrimination ratio and signal-to-noise ratio and compared the performance with a conventional segmented phase-only filter. For carrying out the study, we have developed a dataset of different distorted objects such as alphabets, digits, vehicles, fighter planes. Through a series of experiments and case studies, we are able to prove that the proposed filter reduces the false alarm rate while increasing recognition capability.

show abstract

Composite filtering strategy for improving distortion invariance in object recognition

Cited by 7 publications

References 36 publications

Weight Quantization Retraining for Sparse and Compressed Spatial Domain Correlation Filters

Weight Quantization Retraining for Sparse and Compressed Spatial Domain Correlation Filters

Hardware design of correlation filters for target detection

Wavelet-modified segmented phase-only filter for in-plane rotated object recognition

Contact Info

Product

Resources

About