Temporal Repetition Detector for Time Series of Spectrally Limited Satellite Imagers

Abstract

This article addresses the problem of estimating scene visibility in time series of satellite images. It focuses on satellites with few spectral bands and high revisit frequency. Our approach exploits the redundancy of information acquired during these revisits. It is based on an unsupervised algorithm that tracks local ground textures across time and detects ruptures caused mainly by opaque clouds and in some cases by haze, cirrus and shadows. Experiments have been carried out on 18 PlanetScope image times series of various locations. These time series come with handmade ground truth labels that are published together with this paper. We compare our results with the Unusable Data Masks (UDM) that Planet provides together with the images, and demonstrate the effectiveness of the proposed method: success rates of 97.78% and 89.36% are reached for the visible and occluded regions classification. This article is related to the following publication: [

A CNN Cloud Detector for Panchromatic Satellite Images

“…In this section we test the resulting mask from M against the SIFT cloud detector [15,16]. The two datasets (single and temporal) from [18] are used.…”

“…Alternatively, the inter-band delay in push-broom satellites allows cloud detection by parallax analysis of the color bands [9,10,11,12,13,14]. In some cases, local descriptors are used to detect changes, hopefully due to clouds, and, in other cases, the clouds' relative altitude is measured with respect to the ground through the interband parallax information [15,16,17]. Certainly, the case of cloud detection in panchromatic optical images has none of aforementioned characteristics and presents a challenge as it contains less information to work with.…”

Using the IPOL Journal for Online Reproducible Research in Remote Sensing

“…The article [18] along with its demo 8 addresses the problem of detecting visually stable areas in time series of satellite images. It can be used to detect cloud-free areas.…”

Ground Visibility in Satellite Optical Time Series Based on A Contrario Local Image Matching

“…Instead of detecting clouds, an alternative approach is to detect ground visibility, that is, the parts of each image where the ground is visible [7,6,16]. This can be done by comparing the corresponding parts of a time series and selecting matching regions.…”

“…There are many fast local image comparison methods, and most achieve contrast invariance by relying on the image gradient orientation, like the celebrated SIFT algorithm [26,33]. An effective ground visibility detection algorithm based on SIFT image descriptors was presented in [7,6].…”