Quantization of the Space of Structural Image Features as a Way to Increase Recognition Performance

Gorokhovatskyi, Volodymyr; Gorokhovatskyi, Oleksii; Yevgenyi, Putyatin; Olena, Peredrii

doi:10.1109/dsmp.2018.8478434

Cited by 3 publications

(5 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, it increased the intersection of areas of image descriptions from different classes. In situations where the etalon was represented by a single image, the recognition accuracy for the matching of all key point features and the proposed method was nearly the same [4,6].…”

Section: Discussionmentioning

confidence: 90%

“…At the same time, the use of the two methods described earlier [3,4] was possible. The first was based on the construction of an integral vector representation for the object O , and the second was based on the summation of the vectors of the specific weights of elements classified, according to the rule of Equation 5, to the nearest of the clusters.…”

Section: Properties Of Structural Image Description Learningmentioning

confidence: 99%

“…Structural image recognition methods, where visual objects on images are described as sets of characteristic features (CF), are widely used due to the high efficiency associated with the practical ability to perform recognition in the complex conditions related to geometric transformations, background influences or even the partial absence of features [1][2][3][4]. The user of the computer vision system independently determines which specific part of the available description is acceptable for making a decision about the class of an object.…”

Section: Introductionmentioning

confidence: 99%

“…The user of the computer vision system independently determines which specific part of the available description is acceptable for making a decision about the class of an object. Successful implementation of structural methods dealing with problems of face recognition, animal recognition and recognition of a number of different icon-type images, such as coats of arms, paintings, logos and brands, are known [4][5][6][7].…”

Section: Introductionmentioning

confidence: 99%

“…A way to increase the effectiveness of the structural recognition methods in terms of speed and, in fact, without reducing the probabilistic characteristics, is by the use of vector quantization in the space of structural dataset features [3,4,6]. Quantization implements an approximation in the space of key point feature descriptors by splitting them into subsets of equivalent elements.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Analysis of Application of Cluster Descriptions in Space of Characteristic Image Features

Gorokhovatskyi

Peredrii

2018

Data

Self Cite

View full text Add to dashboard Cite

In this paper, we propose an investigation of the properties of structural image recognition methods in the cluster space of characteristic features. Recognition, which is based on key point descriptors like SIFT (Scale-invariant Feature Transform), SURF (Speeded Up Robust Features), ORB (Oriented FAST and Rotated BRIEF), etc., often relating to the search for corresponding descriptor values between an input image and all etalon images, which require many operations and time. Recognition of the previously quantized (clustered) sets of descriptor features is described. Clustering is performed across the complete set of etalon image descriptors and followed by screening, which allows for representation of each etalon image in vector form as a distribution of clusters. Due to such representations, the number of computation and comparison procedures, which are the core of the recognition process, might be reduced tens of times. Respectively, the preprocessing stage takes additional time for clustering. The implementation of the proposed approach was tested on the Leeds Butterfly dataset. The dependence of cluster amount on recognition performance and processing time was investigated. It was proven that recognition may be performed up to nine times faster with only a moderate decrease in quality recognition compared to searching for correspondences between all existing descriptors in etalon images and input one without quantization.

show abstract

Section: Discussionmentioning

confidence: 90%

Section: Properties Of Structural Image Description Learningmentioning

confidence: 99%