245 
classification precision is 86.20%, and the Kappa coefficient is 
0.8275 o According to the MLC results, this paper utilizes the 
Spot Class Statistic and Judge Function to judge the attribute of 
each spot, and makes the attribute matching between the 
classified and judged spot units and the object codes, then 
detects the changed spots, marks them in high-light (In the 
processing of detection, the water is not considered).Fig 3 is the 
change detection result chart in land use. In this chart, 114 
refers to the dry farmland, 111 refers to the water surface, 115 
refers to vegetable plot, and 203 refers to residential area. 
diuoY inalbox 
ytraiavinU lf>3: 
arii 1o crungoY 
teomriosT §;nn< 
water □ vegetable plot m glebe |/'/) } residential area m paddy field 
Figure 2 overlays the MLC results Figure 3 the change detection result chart in land use 
and the present land use map 
vt balcluolso ariT .lotifb 
v\ -.1 
Sample 
vegetable plot 
residential area 
paddy field 
glebe 
water 
total number 
vegetable plot 
94 
13 
7 
9 
0 
Oft) :••>!> boriloir: ' # tf* i 
123 
residential area 
6 
80 
4 
0 
20 
; aoitibbA .euudhm a 
110 
paddy field 
0 
7 
89 
1 
2 
> , .fflgl>rni ni 8C‘igri6 33il0bl'jni 
99 
iJB3 3rft ?.3lcqi3im>q riurit .ruub 
glebe 
0 
0 
0 
90 
0 
90 
.allunoi tofiaom y36 : i3 rigiri 
water 
0 
0 
0 
0 
78 
78 
SUM 
100 
100 
100 
100 
100 
431 
— 
Table 1 the confusion matrix of MLC results 
4.2 Precision Analysis and Conclusion 
(1) Detection precision: Table 2 is the results of the 
change detection in land use. Monitoring region total area is 
29437.7717 acres, correct monitor region total area is 
28171.9481 acres, misjudgment total area is 1265.8236 acres, 
the number of the misjudgment spots is 21, in which the 
number of the omitted spots is 13, the total omitted area is 
396.4643 acres, monitoring result precision is 95.70%. 
(2) Advantages: has the higher monitoring precision in larger 
regions; obtains the better classified effect to the vegetable and 
the residential area, and overcomes the similarity of texture 
features in both. 
(3) Major effect factors: © the precisions of image 
classification results; (2) different objects have the same 
spectrum (3) the influence of soil spectral feature; © the 
different spectral features of crops before and after harvest; (5) 
the judgment standard based on statistical principles contains 
certain errors, and causes lower monitoring precision in small 
regions. 
m bnc.J gnrenai? alomafl no /bui2 .6002 ,nu (gosi /. iJ 
Color 
Before 
Detection 
After Detection 
Red vector graph unchanged spot 
Spot Area 
(Acre) 
27222.102 
Spot Number 
285 
Misjudged Spot 
Number 
I noi 
Misjudged 
Area 
(Acre) 
396.4643 
isriO 
drift 
Yellow vegetable plot residential area 976.5220 5 
0 
0