Table 3.
AREs for three clinical metrics of three types of tactile sensors after eight denoising methods.
| Butterworth [19] | Wavelet [20] | EMD [22] | VMD [24] | MA [26] | RPCA [27] | WRPCA [28] | cc-DWRPCA [13] | ANOVAa | ||
|---|---|---|---|---|---|---|---|---|---|---|
| 3 × 4 | pAIx | 0.095 ± 0.039 | 0.210 ± 0.067 | 0.152 ± 0.061 | 0.096 ± 0.038 | 0.138 ± 0.058 | 0.083 ± 0.032 | 0.084 ± 0.031 | 0.083 ± 0.031 | p < 0.05 |
| h3/h1 | 0.197 ± 0.094 | 0.419 ± 0.108 | 0.284 ± 0.106 | 0.187 ± 0.093 | 0.230 ± 0.096 | 0.178 ± 0.098 | 0.178 ± 0.095 | 0.177 ± 0.093 | p < 0.05 | |
| h4/h1 | 0.116 ± 0.043 | 0.167 ± 0.062 | 0.169 ± 0.056 | 0.115 ± 0.043 | 0.122 ± 0.049 | 0.093 ± 0.038 | 0.093 ± 0.038 | 0.093 ± 0.038 | p < 0.05 | |
| 4 × 6 | PAIx | 0.225 ± 0.084 | 0.341 ± 0.115 | 0.262 ± 0.074 | 0.224 ± 0.084 | 0.280 ± 0.102 | 0.199 ± 0.069 | 0.197 ± 0.068 | 0.199 ± 0.069 | p < 0.05 |
| h3/h1 | 0.496 ± 0.158 | 0.762 ± 0.189 | 0.554 ± 0.139 | 0.477 ± 0.158 | 0.541 ± 0.160 | 0.451 ± 0.158 | 0.448 ± 0.157 | 0.448 ± 0.157 | p < 0.05 | |
| h4/h1 | 0.324 ± 0.169 | 0.366 ± 0.175 | 0.374 ± 0.160 | 0.319 ± 0.165 | 0.331 ± 0.164 | 0.254 ± 0.101 | 0.248 ± 0.104 | 0.247 ± 0.105 | p < 0.05 | |
| 5 × 5−1 | pAIx | 0.045 ± 0.019 | 0.181 ± 0.069 | 0.292 ± 0.163 | 0.062 ± 0.037 | 0.089 ± 0.035 | 0.057 ± 0.018 | 0.056 ± 0.016 | 0.057 ± 0.019 | p < 0.05 |
| h3/h1 | 0.089 ± 0.039 | 0.341 ± 0.072 | 0.476 ± 0.258 | 0.131 ± 0.119 | 0.120 ± 0.048 | 0.084 ± 0.045 | 0.084 ± 0.043 | 0.083 ± 0.042 | p < 0.05 | |
| h4/h1 | 0.064 ± 0.027 | 0.138 ± 0.045 | 0.242 ± 0.086 | 0.101 ± 0.096 | 0.071 ± 0.027 | 0.046 ± 0.019 | 0.045 ± 0.018 | 0.045 ± 0.017 | p < 0.05 |
The one-way ANOVA results indicate the significant difference among AREs of the eight algorithms.