Home Articles Biomass estimation by the stereophonic image analysis

Biomass estimation by the stereophonic image analysis

ACRS 2000

Poster Session 1

Biomass Estimation by The Stereophonic Image Analysis

Hiroyuki Masubuchi, Koji Kajiwara, Yoshiaki Honda.

Center for Environment Remote Sensing (CEReS), Chiba University

1-33, Yayoi-cho, Inage-ku, Chiba, 263-8522 Japan

Tel: (81)-43-290-3845 Fax: (81)-43-290-3857

E-mail:[email protected]

Keywords: Stereophonic image, Biomass, Grass Height

1.Intoroduction

These days, an environmental problem becomes serious. It is important to grasp a change in the plant in the earth scale. The change in the plant is a great influence an environment. Other side, a detailed remote sensing is becoming available because of a satellite’s ability improved more high resolution and many channels. Our laboratory has been analyzing satellite data for the estimate biomass in processing. And we have been getting real biomass data by the observation on any places. The methods of observation are biomass measurement and other measurement using car, tower, RC Helicopter. In the biomass measurement, we measure the dry weight, grass height, spectral reflectance in detail on the small cell 1 square meter. In the other measurements, we measure the spectral reflectance and picture on the large area. We want to measure any samples and quickly. But on the biomass measurement, we have to crop grass for a long time. It is heavy work. And we can’t crop grass, if we measure it for long run. It needs method to know biomass by the non-contact.


2.Objective

A study of method to estimate biomass using the stereophonic images analysis.


3.Stereophonic Images Measurement System

3D scanner and Stereophonic image are available to measure the height of object by non-contact. 3D scanner can measure correct height and dimension detail. However it needs long time for measuring and it is not easy to move. Other way, the height accuracy using the stereophonic image is lower than 3D scanner. But system is simple and mobility. And it is practicable other observation. Therefore stereophonic image was selected.

Figure 3.1 shows the Stereophonic image measurement system. It has two digital cameras, spectrometer, white board for the reference of spectrometer.

Digital camera is CAMEDIAC-3030ZOOM, OLYMPAS has 3 million pixels CCD and it has non-compress image storage mode (TIFF). It will able to get high-resolution image.


Fig.3.1 Stereophonic image measurement system

Equation 3.1 means the accuracy of height. L is height of camera, is focal length, is pitch of pixel on CCD, B is camera distance.

The accuracy of height is lower 2mm is better because of low grass at Mongolia. So for, camera distance is 1050mm.


Fig.3.2 images



4.Measurement


4.1.Measurement Work Sequence

Biomass measurement work sequence has 5steps. At first, it makes the cells on ground. Cell size is 1 square meter. Measurement target is inside a cell. 2nd, the spectral reflectance and stereo image are gotten, and grass is measured height. 3rd, All grass is cropped. 4th, grass is dried into an oven for 72 hours. At last, measure the dry grass weight.


4.2.Observation Site

In June 2000, we had observation in North America then in august we had observation in Mongolia.


In North America

We visited to Tucson-Arizona, LusCruces-New Mexico, Conza-Kansas and Missoula-Montana. The observation data has a wide variety of environment.


Fig.4.1 Observation site in North America


Fig.4.2 Measuring in North America

ACRS 2000

Poster Session 1

Biomass Estimation by The Stereophonic Image Analysis


In Mongolia

We had observation in Mandalgob for 280km south of Ulanbator.


Fig.4.3 Observation in Mongolia


Fig.4.4 Measuring in Mongolia



5.Analysis


5.1.Estimate grass height

Figure 5.1 shows process flow for the estimate grass height. At first, it compensates distortion of images. Then height map is calculated by the parallax error using a based mapping method of stereo matching. In addition, it makes the vegetation cover image by the extract vegetation from compensated image using principal component analysis. The height map part of vegetation cover and the height map part of ground (it is the other vegetation cover) are made with height map and vegetation cover image. The ground height map has many holes by the part of vegetation cover. So, there are estimate from pixels around the hole. The height map of vegetation is made difference in height between vegetation and ground. Figure 5.2-5.8 show the images are made by each process.


Fig.5.1 Process Flow



5.2 Estimate Biomass

The equation of translation from volume to biomass is liner function as weight density (k) is constant.

Biomass=volume*k (5.1)

Figure 5.9 shows that the vegetation volume calculated from the height map of vegetation. Table 5.1 shows the compared result and true value.


Fig.5.9 Estimated Volume-Biomass

Table5.1 Result

Cell No.
Volume [cm3]
Biomass [g]
Estimated Biomass [g]
Error [%]

JO01
7972
109.33
215.244
96.9

JO02
2975
82.08
80.325
2.1

JO03
5076
108.03
137.052
26.9

JO04
2361.97
100.74
63.77319
36.7

JO05
1515.7
78.54
40.9239
47.9

B00
180.5
9.92
4.8735
50.9

B01
253
10.08
6.831
32.2

B02
207
11.86
5.589
52.9

B03
341
10
9.207
7.9

B04
420
8.58
11.34
32.2

ACRS 2000

Poster Session 1

Biomass Estimation by The Stereophonic Image Analysis


6.Conclusion

We estimate biomass from stereophonic image. But their results have error. As for main cause of error is equation of translation volume to biomass. Weight density is needs to fix better using many processed data and need to consider grass height. In addition, it is necessary for inspect height map and vegetation cover image because error is multiplied. It will become better precision of the estimate biomass by improved algorithm.

This work has been supported by CREST (Core Research for Evolutional Science and Technology) of Japan Science and Technology Corporation (JST). And we are very grateful to many staffs at US and Mongolia

Fig.5.2 Digital Camera Image (L : Left R : Right)

Fig.5.3 Compensated Image

Fig.5.4 Vegetation Cover Image

Fig.5.5 Height map

Fig.5.6 Height map of Estimated Ground

Fig.5.7 Height map part of Vegetation cover

Fig.5.8 Height map of vegetation (Result)