Research Article
AData-DrivenandBiologicallyInspiredPreprocessingSchemeto
Improve Visual Object Recognition
Zahra Sadat Shariatmadar andKarimFaez
Electrical Engineering Department, Amirkabir University of Technology, Tehran 15914, Iran
Correspondence should be addressed to Karim Faez; kfaez@aut.ac.ir
Received 14 October 2020; Revised 28 December 2020; Accepted 20 January 2021; Published 29 January 2021
Academic Editor: Akbar S. Namin
Copyright © 2021 Zahra Sadat Shariatmadar and Karim Faez. is is an open access article distributed under the Creative
Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the
original work is properly cited.
Autonomous object recognition in images is one of the most critical topics in security and commercial applications. Due to recent
advances in visual neuroscience, the researchers tend to extend biologically plausible schemes to improve the accuracy of object
recognition. Preprocessing is one part of the visual recognition system that has received much less attention. In this paper, we
propose a new, simple, and biologically inspired pre processing technique by using the data-driven mechanism of visual attention.
In this part, the responses of Retinal Ganglion Cells (RGCs) are simulated. After obtaining these responses, an efficient threshold is
selected. en, the points of the raw image with the most information are extracted according to it. en, the new images with
these points are created, and finally, by combining these images with entropy coefficients, the most salient object is located. After
extracting appropriate features, the classifier categorizes the initial image into one of the predefined object categories. Our system
was evaluated on the Caltech-101 dataset. Experimental results demonstrate the efficacy and effectiveness of this novel method
of preprocessing.
1.Introduction
One of the challenges in the field of artificial intelligence is
object recognition. e objective of this process is to
classify an object into one of the predefined categories.
ere are various challenges in this field, such as cluttered
and noisy background or objects under different illumi-
nation and contrast environments. Human beings can
detectandclassifyobjectswithoutanyeffortinashorttime.
Researchers believe that the recognition system is closer to
the human visual system will be better. In other words,
numerous studies [1–3] have shown that inspired by the
human visual system, the recognition system can be
designed with relatively high accuracy. According to the
recent advances in visual neuroscience, the researchers
tend to develop biologically plausible algorithms to im-
prove the accuracy of the object recognition system. Object
recognition considerably relies on image representation,
for which, in this paper, a novel biologically inspired model
is presented for this stage. Among image representation
models, bag-of-words (BoW) representation [4] has been
generally employed because it is robust to object scale and
translation changes. ree different modules of BoW
models are extraction, coding, and pooling of different
features. K-means clustering, which is applied for feature
coding, will cause severe information loss because of the
hard assignment of each feature to the nearest cluster
center. So, soft k-means [5] and sparse coding [6] proce-
dures are presented to overcome this problem.
Sparse coding-based methods are broadly used since
they have fewer parameters and more reliable performance
than soft k-means. Some different sparse coding-related
feature coding techniques [6–8] are offered and obtain the
best achievement for image presentation. During the sparse
coding-based strategies, the image is represented by a vector
of sparse codes matching the features in the individual image
within the feature pooling module.
In the BoW model, the whole image is the pooling area,
and therefore, the spatial information may be lost. Such data
can significantly affect recognition accuracy.
Hindawi
Computational Intelligence and Neuroscience
Volume 2021, Article ID 6699335, 13 pages
https://doi.org/10.1155/2021/6699335