Image Based Facial Expression Recognition Using Local Neighborhood Difference Binary Patterns


Downloads per month over past year

Singh, S and Saurav , S and Yadav , M (2018) Image Based Facial Expression Recognition Using Local Neighborhood Difference Binary Patterns. In: 3rd International Confernece on Compute Vision Image Processing, September 29 - October 01, 2018, Jabalpur . (Submitted)

PDF - Submitted Version
Download (5Mb) | Preview


Abstract. Recent l y, automatic facial expression recognition (FER) has gained eno rmous in­ terest among the computer vision researchers because of their potential deployment in a number of industrial, consumer, automobile and societal applications. There are a number of tech niqu es avail abl e in the literature for FE R, among the m man y appearance-based methods such as local binaiy patte rn (LBP), local directional pattern (LDP), local ternary pattern (LTP), gradient local te rnmy pattern (GLT P) and improv ed local te rna ry pattern (lGLTP) have been shown to be very efficient and accurate. In this paper, we have proposed a new descrip tor called Local Neighborhood Dit1erence Binary Pattern (LNDBP). This new descriptor is motivated by the recent success of local neighbo rhood d iffe rence pattern (LNDP) which have been proven to be very effective in image retrieval. The basic characteristics of LNDP as compared to the tradi­ tional LBP is that it generates binary patterns based on a mutual rel ations hi p of all neighboring pixels. Ho wever, in the case of LBP the mutu al rela tio ns h i p does not exist, here only the neigh­ boring pixels are compared with the central pixel to generate the binary pattern and hence there is a los s ofinfonnation which is well captured by LNDP. Therefore, in order to ut ili ze the bene­ fit of both LNDP and LBP, we have proposed LNDBP. We have also employed a dimensionali­ ty reduction technique to reduce the dimens io n of the LNDBP features. The reduced featur es are then classified using kernel extreme learning machine (K-ELM) classifier. In order to vali­ date the performance of the proposed method , experiments have been conducted on two differ­ ent FER datasets. The performance has been observed using we ll- known evaluation mea sures s uch as accuracy, precision, recall, and FI-Score. The proposed method has been compared wi th some state-ot the-art works available in l iterat ure and found to be ve ry effective and accu­ rate.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: Keywords: Facial Expression Recogn i tio n (FER), Local Neighborhood Dit1er­ ence Pattern (LNDP), Principal Component Analysis (PCA), Kernel Extreme Leaming Machine (K-ELM).
Subjects: Electronic Systems > Digital Systems
Divisions: Electronic Systems
Depositing User: Mr. Rabin Chatterjee
Date Deposited: 27 Jul 2021 09:49
Last Modified: 27 Jul 2021 09:49

Actions (login required)

View Item View Item