Geometric Accuracy Assessment and Correction of Imagery from Chinese Earth Observation Satellites (HJ-1 A/B, CBERS-02C and ZY-3)

Abstract

ABSTRACT:The Chinese satellites HJ-1 A/B, CBERS-02C and ZY-3 have been recently launched and are considered as the main space platforms on orbit to acquire optical images for monitoring the Earth for various applications in China. The commercially distributed products (Level 1 or 2) of those satellites usually lack sufficient information (about platform, sensor and ephemeris) that is the key to geometrically correct the acquired images. It is therefore always a challenging issue and the first step to assess the geometric accuracy, which is a key part of qualities in spatial data, of the images from those satellites before generation of geometrically accurate image products. This paper first describes an operational methodology to assess the geometric accuracy of those satellite images. The methodology automatically collects dense and spatially well distributed ground control points (GCP) against reference imagery and then fits those GCPs to the given geometric math model. The geometric accuracy of an image can then be assessed from the overall fitness of those GCPs and their distribution of geometric errors along and across track. The residual mean square (RMS) parameter is used to indicate the degree of overall fitness of the GCPs to the photogrammetric system. The distribution of geometric errors may be random or approximated by a second or higher order polynomial functions; the latter case is generally considered as a systematic error that was not removed completely in the Level 1 or 2 data product. In order to draw solid conclusions, a significant number of samples are selected for each of those satellites by taking variations of landscapes into consideration. The assessment experiments demonstrate that the accuracy of HJ-1 A/B is often very poor, that of CBERS-02C is better than the situation of HJ-1 A/B but records poor accuracy for most samples, and that of ZY-3 is the best among all satellites under investigation and has few samples with poor accuracy. According to the assessment results, this paper suggests an operational correction methodology to improve the accuracy for those satellites, particularly for the HJ-1 A/B and CBERS-02C. Operational production proves that the proposed correction methodology is capable of achieving much higher accuracy than traditional ones and the achieved accuracy meets high standard product requirements for such applications as mapping.

Building high-performance system for processing a daily large volume of Chinese satellites imagery

“…Experiments show than the operational correction methodology works well for those satellites include HJ-1 A/B, ZY-1-02C and ZY-3 [56]. Particularly, the new operational methodology improves the accuracy significantly of the ortho products from the HJ-1 A/B satellites in comparison to the results by the traditional methodology.…”

“…Another paper [56] written by the same authors of this paper proposes an operational methodology to assess geometric accuracy before correction. Assessment of the quality of satellite image data, particularly the geometric accuracy, is always the fundamental step toward finding solutions and then processing the data.…”

“…When the given math model is not suitable for correction after assessment, the paper [56] proposes an operational correction methodology. In the operational correction methodology, a new RPC model is computed from a well distributed GCPs automatically collected from reference data.…”

“…Particularly, the HJ CCD scenes have two rounds of GCP Collection so as to compute a new RPC using the dense GCPs collected automatically in the second round. The first round GCP collection serves to correct large spatial shift between raw images and reference imagery, which can improve successful rate of image matching in the second round of GCP Collection [56].…”

Geospatial Analysis of Horizontal and Vertical Urban Expansion Using Multi-Spatial Resolution Data: A Case Study of Surabaya, Indonesia

“…The geometric correction using 55 GCPs from the base map was performed. The RMSe of 3.68 m, which is equivalent to less than a half pixel, was acceptable [42,43]. The same GCPs were applied for the geometric correction of the ALOS DSM.…”