Limits...
A Multispectral Image Creating Method for a New Airborne Four-Camera System with Different Bandpass Filters.

Li H, Zhang A, Hu S - Sensors (Basel) (2015)

Bottom Line: For this multispectral system, an automatic multispectral data composing method was proposed.For the difficult registration problem between visible band images and near-infrared band images in cases lacking manmade objects, we presented an effective method based on the structural characteristics of the system.Experiments show that our method can acquire high quality multispectral images and the band-to-band alignment error of the composed multiple spectral images is less than 2.5 pixels.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of 3D Information Acquisition and Application of Ministry, Capital Normal University, Beijing 100048, China. lihanlun@126.com.

ABSTRACT
This paper describes an airborne high resolution four-camera multispectral system which mainly consists of four identical monochrome cameras equipped with four interchangeable bandpass filters. For this multispectral system, an automatic multispectral data composing method was proposed. The homography registration model was chosen, and the scale-invariant feature transform (SIFT) and random sample consensus (RANSAC) were used to generate matching points. For the difficult registration problem between visible band images and near-infrared band images in cases lacking manmade objects, we presented an effective method based on the structural characteristics of the system. Experiments show that our method can acquire high quality multispectral images and the band-to-band alignment error of the composed multiple spectral images is less than 2.5 pixels.

No MeSH data available.


The CIR composite of the four-band images; (a–c) respectively the multispectral images obtained by using the methods using the first-order polynomial, second-order polynomial and homography model and using the RANSAC directly; (d) the multispectral images obtained by using our method.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4541943&req=5

sensors-15-17453-f016: The CIR composite of the four-band images; (a–c) respectively the multispectral images obtained by using the methods using the first-order polynomial, second-order polynomial and homography model and using the RANSAC directly; (d) the multispectral images obtained by using our method.

Mentions: Compared with Figure 4, nearly all of the ground objects shown in Figure 6 are the grass except a road. The four methods in the previous section are used to compose one four-band multispectral image. Figure 14 depicts the true-color composite and CIR composite. Figure 14a shows the true-color composite of the unregistered multispectral image, and Figure 14b is its enlarged partial region. An obvious dislocation can be seen in Figure 14a,b. Figure 14c shows the registered true-color composite, and Figure 14d is its enlarged partial region. The dislocation is missing. Because of the lack of artificial objects in these images, there is a great difference between visible SIFT features and infrared SIFT features, and the correct rate of initial matching point pairs is significantly lower than 50%. The first three methods use the RANSAC directly; therefore they cannot get a correct infrared band in the four-band multispectral image, as shown in the first three severely distorted images of Figure 15. However, the fourth method uses the rejecting false matches method mentioned in Section 3 to remove most false matches first for promoting the correct rate, and then uses the RANSAC, so it can get a correct infrared band, as shown in the fourth picture of Figure 15. So, the first three methods cannot get a correct CIR composite, as shown in the first three images of Figure 16, but the fourth can, as shown in the fourth image of Figure 16. Table 3, quantitative evaluation of registration error, shows that the errors of the third method and the fourth method are the same, and a little better than the first and second method at column G-B and G-R. At the G-IR column, the first three methods cannot obtain correct results, but the fourth method can get correct result; and its total error, about 2.4 pixels, is still very low.


A Multispectral Image Creating Method for a New Airborne Four-Camera System with Different Bandpass Filters.

Li H, Zhang A, Hu S - Sensors (Basel) (2015)

The CIR composite of the four-band images; (a–c) respectively the multispectral images obtained by using the methods using the first-order polynomial, second-order polynomial and homography model and using the RANSAC directly; (d) the multispectral images obtained by using our method.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4541943&req=5

sensors-15-17453-f016: The CIR composite of the four-band images; (a–c) respectively the multispectral images obtained by using the methods using the first-order polynomial, second-order polynomial and homography model and using the RANSAC directly; (d) the multispectral images obtained by using our method.
Mentions: Compared with Figure 4, nearly all of the ground objects shown in Figure 6 are the grass except a road. The four methods in the previous section are used to compose one four-band multispectral image. Figure 14 depicts the true-color composite and CIR composite. Figure 14a shows the true-color composite of the unregistered multispectral image, and Figure 14b is its enlarged partial region. An obvious dislocation can be seen in Figure 14a,b. Figure 14c shows the registered true-color composite, and Figure 14d is its enlarged partial region. The dislocation is missing. Because of the lack of artificial objects in these images, there is a great difference between visible SIFT features and infrared SIFT features, and the correct rate of initial matching point pairs is significantly lower than 50%. The first three methods use the RANSAC directly; therefore they cannot get a correct infrared band in the four-band multispectral image, as shown in the first three severely distorted images of Figure 15. However, the fourth method uses the rejecting false matches method mentioned in Section 3 to remove most false matches first for promoting the correct rate, and then uses the RANSAC, so it can get a correct infrared band, as shown in the fourth picture of Figure 15. So, the first three methods cannot get a correct CIR composite, as shown in the first three images of Figure 16, but the fourth can, as shown in the fourth image of Figure 16. Table 3, quantitative evaluation of registration error, shows that the errors of the third method and the fourth method are the same, and a little better than the first and second method at column G-B and G-R. At the G-IR column, the first three methods cannot obtain correct results, but the fourth method can get correct result; and its total error, about 2.4 pixels, is still very low.

Bottom Line: For this multispectral system, an automatic multispectral data composing method was proposed.For the difficult registration problem between visible band images and near-infrared band images in cases lacking manmade objects, we presented an effective method based on the structural characteristics of the system.Experiments show that our method can acquire high quality multispectral images and the band-to-band alignment error of the composed multiple spectral images is less than 2.5 pixels.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of 3D Information Acquisition and Application of Ministry, Capital Normal University, Beijing 100048, China. lihanlun@126.com.

ABSTRACT
This paper describes an airborne high resolution four-camera multispectral system which mainly consists of four identical monochrome cameras equipped with four interchangeable bandpass filters. For this multispectral system, an automatic multispectral data composing method was proposed. The homography registration model was chosen, and the scale-invariant feature transform (SIFT) and random sample consensus (RANSAC) were used to generate matching points. For the difficult registration problem between visible band images and near-infrared band images in cases lacking manmade objects, we presented an effective method based on the structural characteristics of the system. Experiments show that our method can acquire high quality multispectral images and the band-to-band alignment error of the composed multiple spectral images is less than 2.5 pixels.

No MeSH data available.