CHAPTER 5
RESULT AND ANALYSIS
5.1 Result
Two images of SAR and landsat8 are classified with automated signature using maximum likelihood algorithm. Fig. 6.9 shown classified map of the SAR image with automated training technique using maximum likelihood method which contains two classes, namely water and non water bodies. Fig.6.6 illustrate the classified map of landsat8 multispectral image which consists of four classes, such as water bodies , vegetation, bare land, buildup & sand area. Fig 6.16 shown the SAR image classification on the basis of ISODATA method.

The water bodies visualized in classified landsat8 images are less than the amount of water bodies shown in the SAR images. The water bodies are depicted in the classified SAR …show more content…
5.15 Maximum likelihood classification accuracy checking.

Fig. 5.16 ISODATA classification accuracy checking.

Table shows confusion matrix for landsat8 and SAR classified images.

Table 5.1 Confusion matrix for automated technique with maximum likelihood classification on Lansat8 image.
Class Reference_1 Reference_2 Reference_3 Reference_4 Ground truth water 46 0 1 0 47
Vegetation 1 49 2 1 53
Buildup & sand 3 0 46 1 50
Bare land 0 1 1 48 50 total 50 50 50 50 200

Table 5.3 Confusion matrix for maximum likelihood classification on Landsat8 image.
Class Reference_1 Reference_2 Reference_3 Reference_4 Ground truth water 43 0 1 0 44
Vegetation 1 47 0 0 48
Buildup & sand 6 3 49 21 79
Bare land 0 0 0 29 29 total 50 50 50 50 200

Confusion matrix for SAR image.

Table 5.5 Confusion matrix for automated technique with maximum likelihood classification
Class Reference_1 Reference_2 Ground truth
Water 46 5 51
Non water bodies 4 45 49
Total 50 50 100

Table 5.7 maximum Confusion matrix for likelihood classification on SAR image
Class Reference_1 Reference_2 Ground truth
Water 47 9 56
Non water bodies 3 41 44
Total 50 50