Automatic Image Self-Enhancement for Multi-Scale Spectral Residual on Low-Resolution Video

Automatic Image Self-Enhancement for Multi-Scale Spectral Residual on Low-Resolution Video

Exposure Fusion Framework, Human Detection, Multi-Scale Spectral Residual

Multi-Scale Spectral Residual technique is used to reduce the search area in an image. However, this technique relies on image salience from the capture device. The aim of this study is to obtain a better search area with image enhancement to detect human objects on low resolution video. Enhanced image uses only pixels in each frame of the video using the Exposure Fusion Framework. The dataset is an artificial video obtained from a room with low resolution CCTV. This study compares the detection results before and after applying
image enhancement on MSR. We are adopting Linear-SVM based human detection with
Histogram of Gradient (HOG) features as a test case. Human detection was evaluated using
precision, recall, f-score rate and validated by leave-one-out cross validation. The results show that enhanced images can improve overall performance by 64.46% compared to the original video in human detection on low resolution video, with an increase in recall of 3.21%

Arwin Halim, Sunaryo Winardi, and Erlina Halim

http://dx.doi.org/10.21609/jiki.v14i1.900

Faculty of Computer Science Universitas Indonesia

2021-02-28

Sri Wahyuni

e-ISSN : 2502-9274 printed ISSN : 2088-7051

PDF

English

Text

Jurnal Ilmu Komputer dan Informasi (Journal of Computer Science and Information)