GISdevelopment.net ---> AARS ---> ACRS 1990 ---> Poster Session Q

A fast image preprocessing system and its software for airborne MSS

Liu Dingsheng, Li Xia, Li Zhirong
Remote sensing Satellite station
Chinese Academy of science


Abstract
A fast image preprocessing system RTPS-1 has been developed as the ground part of a Chinese airborne practical operational system this system includes two subsystems front end (SHPS) and back end (GIPS)

In the front-end hardware modules was used to do the input from optical disk format transformation B & W quick look rough correction and the productions of tape and diskette.

The back end is a general image processing system with high resolution color display and perfect image processing software besides airborne image preprocessing software package has been developed these software were checked by real data of airborne MSS the data can be transferred between front end and back end by special DMA interface the thought put of system is about 60 scan line / second.

Introduction
Due to the large scanning angle of sensor attitude change of airplane and the atmosphere effect the image preprocessing of airborne multi spectral scanner data is more difficult than the space satellite data in order to complete Chinese air borne practical operational system this RTPS fast image preprocessing in large MSS images rapidly as an specified primary purpose of RTPS-1 was to preprocess in large MSS images rapidly as an operational system to supply the data products for users however it performs image processing and analysis functions as well and can easily be upgraded to expand its function later the system takes the advantages of the technology of firmware and modular and the parallel processing of multi micro processing for the high speed processing and mass production .

To develop a whole set a satisfied process for the air borne image preprocessing was needed in this paper a set of methods was suggested for the systematic preprocessing of air borne MSS data these methods are selected and summarized from several successful methods to make a whole set soft ware package for air borne MSS data.

System Configuration
The whole system configuration of RTPS-1 shown in figure -1 it is consisted of two independent subsystems but with some relation between them --- the front end and back end the SHPS is a hardware-oriented subsystem with high modular design and multi bus. it was made up of special hardware modules multi micro processors and some peripheral interfaces the high speed was obtained by special real time hardware and the parallel processing of multi processors.


The programmable microprocessor and relative independent of each module make the SHPS flexibility the master module using M68000 CPU controls the operation of the various modules and supervises calculates some parameters as well the high speed module of preprocessing is a key module in the front end sub system it performs high speed data processing such as radiometric statistics and calibration rough geometric correction and image enhancement the processed data then send into tape recorder diskette or the back end.

The back end is a general image processing system composed of Micro VAX II computer high resolution color display and some general peripherals the GIPS processed the data come from front end by using of a large image processing software package the GIPS implemented precise geometric correction mosaic and other functions after that the processed data were out put to the tape or returned to the front end.

High-speed module of preprocessing
This module is composed of two boards shown in figure 2 it undertakes the format transformation radiometric geometric correction.


  1. Format Transformation

    The function of format transformation was to change the original data recording format standard for later preprocessing a hardware transformer was designed which could be rasily changed by program . Each pixel in the input data system was assigned a pointer through a programmable look up table the pointer mapped that pixel toa corresponding positions in the output stream using this method any input format with in 64 bases could be transformed in to any out put format the change rate was 5 m bytes /sec.

  2. Rough Geometric Correction

    Due to the large variation of airborne flighting attitude the correction of the geometric distortion was realized by a high speed programmable unit the core of the unit is TI 32020 high speed signal processor at this time the capability tangent or rolling correction was above 100 k pixel.

  3. The Radiometric Correction

    By the special interface between board FCC and RCC the statistics and correction of radiometric on image pixel in column was realized by hardware the FCC transfers the data and its pointer of positions to the RCC directly then RCC do the statistical and correction works the throughout was see continuously.
Application of optical disk for storage.
The advantages of OD comparing with high density digital recorder is a random reading and reliability the problem of use OD and a data storage to is need developing the special control interface software of river driver and defining data arrangement the continuous recording format is used for data store of each flight line for easy read the position of any flight line a index area is established the interface is a intelligent on which a special circuit was designed to cheek the scan line for existing out the image data high speedly.

Airborne image preprocessing methods.
All of the change of attitude scanner angle electronic angle and electrical and optical system and the atmosphere effect made the degradation based on the real fighting data and several successfully method a complete process of preprocessing airborne scanning data is suggested the flow chart is shown in figure 3 the process divided in to three steps radiometric correction geometric correction and image enhancement.
  1. Normalization of image gray level.


    For easy comparison of different image statistics method is used to consider the effect of absorption and scatterbrain each pixel calculates its means and standard deviation values then its calibrated value is.


    i-line direction ( 1-N1) , j-column direction ( 1-N2)
    M(j) , SD (j) ---- the mean and standard deviation values in column direction
    M,D, mean standard deviation of exceptive value
    X, X' original corrected image value.

  2. Tangent correction

    The large geometric distortion in the edge of image was introduced buy the large airborne scanning angle this distortion can be compensated by the method of tangent correction the principal of the method is resample the image pixel along with the scanning line according to the pixel resolution at the middle point of scan line .

  3. Correction by modulation transfer function (MTF)

    The MTF represents the characters of total scanner's optical and electronic system. The different methods can determine its value a experienced and analytical method is proposed.


    F (u,v), f(x,y) ---- the processed function in the frequency and space domain.
    MTi, MTs --- ideal and supposed MTF of the system
    rs, rp ----- original processed image value.

  4. Geometric correction

    Using the system parameter such as flight height speed and attitude value to correct the geometric distortion is proposed but it will fail when the distortion is large then the precise geometric correction using the ground control points is used.

  5. Image enhancement

    when the correlation between each band is high the color picture become similar as black the enhancement of K-L transformation is very another method is use of the space of luminance but and saturation spaces to enhances the pictures the then return to the RGB space.

  6. Results of experiment

    Using the flight data fine multi spectral scanner processed the images according to the above methods the area is in the western china there are 8 bands and 6000 m highs with 6 mrd IFOV the original images was blurred and shifting among the each bands afetr processing of normalization and tangent the picture became clear and the pixel number from 256 to 396 in each line specially the MTF correction highly increases the quality of image due to high correlation among the FIMS band K-L and LHS transformations are used the results was satisfactory.
Conclusion
By using the multi micro processors and firmware technology fast preprocessing of air borne MSS data was realized in the RTPS -1 system of preprocessing methods was cheek by low quality FIRM data the results was satisfied the proposed methods should be recommended to other airborne MSS data.

Reference
  • B.J Baillie 10th Canadian Symposium of Remote Sensing May 1986