171 
THE ELIMINATE INCONSISTENT ALGORITHM AND ENHANCEMENT OF LARGE 
PLANE ARRAY CCD IMAGES 
GENG Lichuan 3 ’ b , WU Yundong 3 ’ * , GENG Zexun a 
a Institute of Surveying and mapping, Information Engineering University, 450052 Zhengzhou, China - 
glch 1982@ 163 .com 
b 66240 Troops, 100042 Beijing, China 
Youth Forum 
KEY WORDS: CCD Image, Digital Aerial Photogrammetry Camera, limage Edge, Image Processing, Modelling 
ABSTRACT: 
Aiming at uneven illumination and low contrast of large plane array CCD images, this paper presents an effective algorithm to 
extract uneven illumination from images and enhance the contrast by using wavelet transform. Based on an analysis of the 
illumination model on a focal plane, emulational images are constructed for the illumination model. Through analysis of the wavelet 
transform coefficients of emulational images, it is found that information of illumination non-uniformity mainly appears in wavelet 
approximate coefficients, and detail information is contained in detail coefficients. Taking advantage of this property, the original 
image is first decomposed with multiresolution wavelet, next, the natural logarithms of approximate coefficients are calculated, and 
an appropriate attenuating operator is applied to implement the non-uniformity correction of approximate coefficients before 
exponentiating correction, linear contrast-stretch is applied to all the approximate coefficients, a nonlinear function is applied to each 
level detail coefficients to accomplish adaptive contrast enhancement, finally, uniform illuminative and contrast enhanced aerial 
images are acquired through reconstruction of images. The experiments show that the wavelet analysis method can achieve a 
satisfying effect of removing the uneven illumination for digital aerial images and keeping well detail image features, contrast 
enhancing aim is also achieved. 
1. INTRODUCTION 
The first large plane colour CCD digital aerial photogrammetry 
camera which made in China uses one 9k* 9k pixels CCD to 
record the panchromatic image and three 2k*2k pixels CCD to 
record the RGB bands. Every CCD corresponding one 
independent optical system, the field of view of each lens is 60 
degree. The image illumination of central projection optical 
lens is inconsistent, central area is brighter than around area, 
and image contrast is also a bit low. Through theory analysis, 
we find that the illumination attenuating model varies according 
to 4 time power cosine function. Several other factors are also 
affecting the image illumination, except for lens structure and 
camera machining techniques. For example, the elimination of 
multi-channel inconsistent of large plane CCD image also 
affects the image illumination attenuation model 1 ' 1 . How to 
eliminate the inconsistent of image illumination and improve 
the image contrast is very necessary. 
MASK dodging algorithm was introduced in [2] and [13] to 
eliminate the illumination inconsistent of aerial images. It’s 
considered that the original image is the superposition of an 
even illuminating image and an uneven illuminating 
background image, the correction of uneven illumination is to 
subtract the background image from the original image. In this 
scheme, the background image must be simulated accurately 
beforehand, but the background image is affected by many 
factors, it’s difficult to construct it. Furthermore, the 
background images of different images are also different. 
The background image can be considered to be low frequency 
signals comparing to the image signal. Wavelet analysis method 
which has multiresolution analysis character is considered in 
this paper. An advantage of wavelet transform is that the time- 
frequency windows vary. In order to isolate signal 
discontinuities, one would like to have some very narrow basis 
functions. At the same time, in order to obtain detailed 
frequency analysis, one would like to have some very long 
basis functions. A way to achieve this is to have short high- 
frequency basis functions and long low-frequency ones. This 
happy medium is exactly what you get with wavelet 
transform 131 . 
After analysis and validation, it is found that the main 
illumination information of the whole image is contained in the 
approximate coefficients, and the detail coefficients contain the 
detail information. Through adjusting the approximate 
coefficients, the illumination un-uniformity of the image can be 
corrected, at the same time the detail information can also be 
enhanced by adjusting the detail coefficients. 
2. OPTICAL LENS VISUAL FIELD ILLUMINATION 
ATTENUATING MODEL ANALYSIS 
s ñ 
As shown in Figure 1, 0 represents out pupil plane, 0 
radius vector of principle point, and n radius vector of non 
principal point, the angle between n 0 and n . Because the 
image plane is lighten by s 0 , so the radiant intensity relation 
between the principle point ( 10 ) and non-principle point (I ) is 
as below 141 : 
* Corresponding author. This is useful to know for communication with the appropriate person in cases with more than one author.