0 引言
1 联合梯度判别与自适应匹配算法
1.1 梯度判别
=
1.2 自适应模型匹配
C(x,y)=A(x,y)÷B(x,y)
1.3 置信度计算与自适应阈值分割
Vc=μPw+(1-μ)Ps
Th=kImax+(1-k)Imean
2 嵌入式系统实现
3 PC端与嵌入式系统的实验测试
3.1 红外数据集的实验测试
表1 实验结果对比表Table 1 Comparison table of experimental results |
| Infrared background | Samples to be tested | Identify samples | Recognition rate/% |
|---|---|---|---|
| Mountain | 200 | 187 | 93.5 |
| Forest | 240 | 225 | 93.75 |
| Hill | 300 | 277 | 92.3 |
3.2 复杂场景下不同算法对比测试
RSC=
GSCR=
FBS=
表2 3种背景下不同算法对比Table 2 Index values of different algorithms under 3 backgrounds |
| Background index | TLLCM | WSLCM | Proposed | |
|---|---|---|---|---|
| Sky | FBS | 1.70 | 4.81 | 6.36 |
| RSC | 0.09 | 0.10 | 31.21 | |
| RSCG | 0.20 | 0.22 | 69.86 | |
| Time/s | 0.34 | 1.12 | 0.20 | |
| Sky- cloud | FBS | 3.42 | 4.09 | 5.41 |
| RSC | 0.62 | 25.12 | 29.32 | |
| RSCG | 0.89 | 38.51 | 42.48 | |
| Time/s | 0.25 | 1.1 | 0.18 | |
| Ground- hill | FBS | 2.87 | 1.87 | 3.84 |
| RSC | 1.85 | 11.94 | 21.35 | |
| RSCG | 1.61 | 10.43 | 18.65 | |
| Time/s | 0.26 | 1.15 | 0.31 | |