1169
AN IMPROVED IHS FUSION METHOD FOR MERGING MULTI-SPECTRAL AND
PANCHROMATIC IMAGES CONSIDERING SENSOR SPECTRAL RESPONSE
Jia Xu, Zequn Guan, Jie Liu
School of Remote Sensing and Information Engineering, Wuhan University, Wuhan, 430079, China
-jiaxu_whu@yahoo.com.cn
KEY WORDS: image processing, sharpening, image fusion, intensity-hue-saturation (IHS) transform, spectral response
ABSTRACT:
While many remote sensing and GIS applications require both the spatial resolution and spectral resolution be high, image fusion, or
in other words, image sharpening, is a useful technique. To date, numerous image fusion techniques have been developed. However,
some undesirable effects such as modified spectral signatures and resolution overinjection are produced. In this paper, a novel
spectral preservation fusion method for remotely sensed images is presented by considering the physical characteristics of sensors. It
is mainly based on the fast intensity-hue-saturation (IHS) transform but improved in two parts: the construction of intensity
component and the injection method of detail information. In the proposed method, the spectral sensitivity of the multispectral and
panchromatic sensors has been taken into account and all the multispectral bands can be fused at the same time. Experiments carried
out on IKONOS, Landsat 7 ETM+ and EO-1 ALI images show that the proposed method can preserve spatial details and minimize
spectral distortion.
1. INTRODUCTION
Most of the newest remote sensing systems, such as Landsat 7,
SPOT, IKONOS, QuickBird, EO-1, ALOS provide sensors with
one high spatial resolution panchromatic (PAN) and several
multispectral (MS) bands simultaneously. Meanwhile, an
increasing number of applications, such as feature detection,
change monitoring, and land cover classification, often demand
the use of images with both high spatial and high spectral
resolution. As a result, the fusion of HRP and LRM images has
become a powerful solution and many image fusion methods
have been proposed over the last two decades (Pohl et al, 1998;
Lau et al, 2000; Wang et al,2005). However, as the physical
spectral characteristic of the sensors are not considered during
the fusion process, some undesirable effects such as modified
spectral signatures and resolution overinjection are produced.
Recently, Otazu et al (2005) has presented a technique which
takes into account the physical electromagnetic spectrum
response of sensors during the fusion process and successfully
applied it to wavelet-based image fusion methods7. Some
image fusion methods which employ the information of sensor
spectral response have already been carried out on IKONOS
images and demonstrated to be effective (Gonzalez-Audicana et
al,2006; Dou et al, 2007; Zhang et al,2007).
In this paper, after analyzing and comparing the radiometric
properties of different sensors, we present a new improved
method based on the fast IHS transform which takes the sensor
spectral response into account. The proposed method minimizes
spectral distortion and is capable of merging all the MS bands at
the same time. To evaluate the performance and efficiency of
the proposed method, experiments are carried out on IKONOS,
Landsat 7 ETM+ and EO-1 ALI images. The proposed method
is compared together with traditional IHS method and three
typical modified IHS methods both visually and quantitatively.
2. SPECTRAL CHARACTERISTICS OF SENSORS
Problems and limitations associated with the available fusion
techniques have been reported by many studies (Zhang,2000).
The most significant problem may be the spectral distortion of
fused images. To understand the influence of senor spectral
response on panchromatic and multispectral image fusion, the
spectral characteristics of different sensors are investigated in
detail.
Satellite
(Sensor)
Spectral range
(pm)
Corresponding MS
Bands
Landsat 7
0.52-0.90
2(G), 3(R), 4(NIR)
IKONOS
0.45-0.90
1(B), 2(G), 3(R),
4(NIR)
Quickbird
0.45-0.90
1(B), 2(G), 3(R),
4(NIR)
SPOT 5
0.48-0.71
1(G), 2(R)
IRS P6
0.50-0.85
1(G), 2(R), 3(NIR)
EOl (ALI)
0.48-0.69
2(B), 3(G), 4(R)
ALOS
0.52-0.77
2(G), 3(R)
Table 1. Spectral ranges of PAN sensors
2.1 Spectral range of panchromatic sensor
A major reason for the significant spectral distortion in image
fusion is the wavelength extension of the new satellite PAN
sensors. Table 1 shows the spectral ranges of different PAN
sensors. It is obvious that their spectral ranges are different. The
spectral ranges of IKONOS, QuickBird and Landsat 7 are wider
than the others and extended from the visible into the near
infrared, which are different from that of SPOT, IRS, ALI and
ALOS. This difference makes the grey value relationship of an
IKONOS, Quickbird or Landsat 7 panchromatic image
significant different from that of other panchromatic images.
For example, as the high reflectivity in near infrared band,