基于比较测量法的LCTF光谱相机光谱反射率重建研究

摘要

基于液晶调谐滤光片(LCTF)的光谱相机由于体积小、功耗低、集成方便,适合多种搭载平台,同时通过光谱反射率来识别目标物,可以有效解决同色异谱的问题,在伪装识别、精细农产品产值估计等方面有很好的应用前景.比较测量法因其能有效排除光照条件和场景背景条件的影响,重建效率较高,在光谱相机的光谱反射率重建中被普遍采用.比较测量法通常采用标准白板作为参考物,但实际应用中,经常会遇到目标场景空间位置不便放置标准白板的情况.针对这种情况,研究采用LCTF光谱相机采集的目标场景中亮度高的颜色作为参考的比较测量法进行目标物的光谱反射率重建来解决上述问题.实验以美国X-Rite的Gretg Macbeth Color Checker(MCC)色卡作为目标物,分别参考亮度高的五个不同色块,运用比较测量法完成红、绿、蓝三个色块的光谱反射率重建.将重建后三个色块的光谱反射率与采用台式分光光度计(X-Rite Color Eye7000A)测量的标准光谱反射比进行比较.结果显示,重建后的红、绿、蓝三个色块的光谱反射率曲线与用分光光度计测量的标准光谱反射率曲线的形状基本相似,每组曲线的均方根误差均小于0.05;重建的蓝色色块的光谱反射率以参考与其光谱反射比接近的蓝绿色重建的效果最好;重建的红色和绿色色块的光谱反射率以除白色外亮度最高的黄色为参考重建的效果最好;二者均比参考标准白板的反射率重建效果好.实验结果说明采用目标场景中亮度高目视效果明显的色块作为参考对象可被用来进行LCTF光谱相机的光谱反射率重建.该方法有效扩大了比较测量法以及液晶调谐光谱成像系统的应用效率.%The spectral camera based on the liquid crystal tunable filter (LCTF) can be used to different platforms due to its advantage of small volume ,low power consumption and convenient integration .In addition ,LCTF spectral camera identifies the color of the target in the scene according to the spectral reflectance characteristic of the target ,which can address the problem of metamerism very well .Thus ,LCTF spectral camera is also applicable to areas of camouflage recognition and estimation of production of fine agricultural products .The comparative measurement method ,which can effectively discount effects of illuminant and background in the scene ,has been widely used to reconstruct the spectral reflectance of the target .The comparative measurement method mainly adopts a standard white surface as a reference .However ,in some scenes ,because of the limitation of the space ,it is very difficult to put in a standard white surface .In order to address this question ,in this paper ,a comparative measurement by referring to the bright targets in the scene obtained by LCTF spectral camera was proposed .In the experiment ,five different surfaces in the Gretg Macbeth Color Checker (MCC) were respectively used as references to construct the spectral reflectance of red ,green and blue surfaces in MCC .The reconstructed spectral reflectance of red ,green and blue surfaces was compared with that measured by a spectrophotometer (X-Rite Color Eye 7000A) .Results showed that reflectance curves reconstructed were basically similar with those measured by the spectrophotometer in the shape;root mean square errors for red , green and blue surfaces were all less than 0.05 .For blue surface ,the reconstructed reflectance was the most accurate by referring to blue-green surface whose reflectance was close to that of blue surface;for red and green surfaces ,the reconstructed reflectance was the most accurate by referring to yellow surface which has the second-largest luminance after the white surface;the above reconstructed reflectance was more accurate than that by referring to the standard white surface .Results indicated that the proposed method by referring to the bright surfaces can be used to construct the spectral reflectance of the target very well in LCTF spectral camera .The proposed method is useful in the expansion of application fields of LCTF spectral camera .