Article Preview
Top1. Introduction
The Landsat 7 scan line corrector (SLC) failed to correct the undersampling of the primary scan mirror on May 31, 2003. On average, about 22% of the total image is missing pixels (Salma et. al., 2017) in each scene. Though NASA has made efforts to correct the SLC malfunction, this has not been successful and the problem prevails. Without the operating SLC, in addition, the areas that have missing pixels are not identical across all multispectral bands. It appears that gaps change positions slightly with spectral bands and produce invalid data in some bands whereas no data in other bands. The non-identical missing pixels in different bands brings out a research problem in trying to identify and fill the data gaps. The Landsat imagery from 1972 to present provides a unique resource for researchers, scientists and users who work in varied domains namely, agriculture, geology, forestry, mapping, regional planning, and global change research. The Enhanced Thematic Mapper Plus was greatly impacted by the failure of the system’s SLC. For this, gapping is a typical phenomenon with remote sensing imagery. Remote sensing imagery could have dynamic and diverse characteristics. Thus a variety of image restoration techniques (Francescopaolo et. al., 2018; Kimuraet. al., 2011; Rareset. al., 2005)and gap filling techniques (Jiaqing et. al., 2019; Giampaolo et. al., 2019; Mohammed, 2013) could be applied to recover the missing regions.
The proposed system consist of three main parts. The first step is to segment the image using Simple Linear Iterative Clustering(SLIC) (Radhakrishnaet. al., 2011) for getting SLIC superpixels and Regionalization with Dynamically constrained agglomerative clustering and partitioning (REDCAP)(Guo, 2007) to get image segments .The second step consist of finding the missing area calculated using Edge Matching Algorithm (Salma et. al., 2017).The third step is used to fill the gaps in the damaged region performed using Accelerated Proximal Gradient Line Algorithm (APGL) (Salma et. al., 2017).
The remaining of the paper is organized as follows: Section 2 gives background on interactive discussion of some related works. Section 3 describes the overall system design, algorithm and detailed description about each module. Section 4 describes about dataset used for the implementation of the proposed framework, results of the experiments conducted, test cases, performance evaluation. Section 5 describes about the results of proposed system with existing system.Section6 discusses about the conclusion.