Embedded platform for local image descriptor based object detection

Kapela, Rafał; Gugała, Karol; Śniatała, Pawel; Świetlicka, Aleksandra; Kolanowski, Krzysztof

doi:10.1016/j.amc.2015.02.029

Cited by 10 publications

(9 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, it is important to compare our proposed cosine angle distance approach for descriptor matching with other implementations in the literature. There are several matching techniques based on different approaches such as calculating the Chi-square distances [9], Sum of Absolute Differences (SAD) [8], and calculating Hamming distances [11].…”

Section: Resultsmentioning

confidence: 99%

“…As the number of Hamming distance calculation cores increases, the number of resources utilization increases. Table 2 shows the utilized resources of our proposed SIFT matching core using cosine angle distance versus other matching methods such as Chi-square distance [9], SAD calculators [8] and Hamming distance [11]. Compared to these recent works [9,8,11] of implementation of matching cores on hardware, our proposed architecture consumes significantly fewer resources with acceptable matching accuracy ( 98%).…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

A fully pipelined FPGA accelerator for scale invariant feature transform keypoint descriptor matching

Daoud

Latif

Jacinto

et al. 2020

Microprocessors and Microsystems

View full text Add to dashboard Cite

The scale invariant feature transform (SIFT) algorithm is considered a classical feature extraction algorithm within the field of computer vision. The SIFT keypoint descriptor matching is a computationally intensive process due to the amount of data consumed. In this paper, we designed a fully pipelined hardware accelerator architecture for the SIFT keypoint descriptor matching. It was implemented and tested on a field programmable gate array (FPGA). The proposed hardware architecture is able to properly handle the memory bandwidth necessary for a fully-pipelined implementation and hit the roofline performance model achieving the potential maximum throughput. The fully pipelined matching architecture was designed based on consine angle distance approach. It was optimized for 16-bit fixed-point operations and implemented on hardware using Xilinx Zynq-based FPGA development board. Our proposed architecture showed a noticeable reduction of area resources compared with its counterparts in the literature maintaining high throughput by alleviating the memory bandwidth restrictions. The results showed reduction in device-resources up to 91% in LUTs and 79% of BRAMs. Our hardware implementation is 15.7× faster than the comparable software approach.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

A fully pipelined FPGA accelerator for scale invariant feature transform keypoint descriptor matching

Daoud

Latif

Jacinto

et al. 2020

Microprocessors and Microsystems

View full text Add to dashboard Cite

show abstract

“…The feature descriptor mentioned above is a binary string that provides increased storage and matching speed. The traditional matching method often utilizes the brute force (BF) [ 29 ] algorithm to match the feature points, which is followed by the RANSAC algorithm to eliminate mismatches. It is effective for normal scenes [ 30 ].…”

Section: Feature Selection and Matching Methodsmentioning

confidence: 99%

A novel image registration approach via combining local features and geometric invariants

Gao

Zhang

et al. 2018

PLoS ONE

View full text Add to dashboard Cite

Image registration is widely used in many fields, but the adaptability of the existing methods is limited. This work proposes a novel image registration method with high precision for various complex applications. In this framework, the registration problem is divided into two stages. First, we detect and describe scale-invariant feature points using modified computer vision-oriented fast and rotated brief (ORB) algorithm, and a simple method to increase the performance of feature points matching is proposed. Second, we develop a new local constraint of rough selection according to the feature distances. Evidence shows that the existing matching techniques based on image features are insufficient for the images with sparse image details. Then, we propose a novel matching algorithm via geometric constraints, and establish local feature descriptions based on geometric invariances for the selected feature points. Subsequently, a new price function is constructed to evaluate the similarities between points and obtain exact matching pairs. Finally, we employ the progressive sample consensus method to remove wrong matches and calculate the space transform parameters. Experimental results on various complex image datasets verify that the proposed method is more robust and significantly reduces the rate of false matches while retaining more high-quality feature points.

show abstract

“…Once the keypoints are detected and features are described, they are then used to match between the test and the reference images, the process called featurematching. Feature matching methods traditionally use the k-nearest neighbor (kNN) [3] or the brute force algorithm [18] to match the features points. Then an outlier detection algorithm such as random sample consensus (RANSAC) [13] is used to eliminate mismatches or outliers.…”

Section: Introductionmentioning

confidence: 99%

Using Local, Contextual, and Deep Convolutional Neural Network Features in Image Registration

Shrestha

2020

Proceedings of the 12th International Conference on Computer Modeling and Simulation

View full text Add to dashboard Cite

Image registration is a well-known problem that arises in many applications in the fields of computer vision, remote sensing, and medical imaging. Many registration methods have been proposed in the literature. However, no single method works well in all kinds of images. In this work, local features and context-based augmented features are used in order to improve the accuracy of the image registration. Furthermore, an attempt has been made to use deep convolutional neural network features on top of those features for further improvement. The paper presents comparative results on image registration with and without feature augmentation and the deep convolutional neural network features. The results from the methods on a widely used benchmark dataset from the University of Oxford confirm improvement in the accuracy of image registration when local and augmented features are used. CCS CONCEPTS• Computing methodologies → Neural networks; Matching.

show abstract

Embedded platform for local image descriptor based object detection

Cited by 10 publications

References 16 publications

A fully pipelined FPGA accelerator for scale invariant feature transform keypoint descriptor matching

A fully pipelined FPGA accelerator for scale invariant feature transform keypoint descriptor matching

A novel image registration approach via combining local features and geometric invariants

Using Local, Contextual, and Deep Convolutional Neural Network Features in Image Registration

Contact Info

Product

Resources

About