基于彩色图像插值算法的胶囊机器人微视觉

Micro-Vision of Capsule Robot Based on Color Image Interpolation Algorithm

针对胶囊机器人小体积对其携带能量的限制,提出了胶囊机器人微视觉技术的思想;直接利用其内部的基于Bayer型彩色过滤阵列的图像传感器采集到的RAW RGB数据来进行传输、存储操作,其他的工作交给后期的计算机工作站完成;对于计算机工作站上图像显示的工作,提出采用一种基于最优化彩色空间转换的多梯度彩色插值算法;该算法首先利用多梯度算子来恢复绿色分量,然后在一个最优化彩色空间内恢复另外两个颜色分量;通过体外环境和模拟体内黑暗液体环境下的实验结果显示,此插值算法相对于传统的插值算法减小了误差,调高了峰值信噪比,改善了图像的边缘模糊现象,并验证了此方法能够满足内窥镜检测技术所需要的图像效果。

For wireless capsule robots carries the limited energy as its small volumn, this paper puts forward the idea of capsule robots u- sing micro vision. It directly uses the RAW RGB data that was collected by Bayer color filter array image sensor inside the capsule robot to be transferred and stored. Other tasks are handed to the host computer. For the work of images showing in the host computer, this paper pro- pose a multi--color gradient image interpolation algorithm based on optimization color space conversion. The algorithm firstly restores the green component using multi--color gradient, and then restores the other two color components in an optimized color space. Finally, experi- mental results in vitro environment and in vivo simulation of dark liquid environment showed that this algorithm is able to reduce the MSE , improve the PSNR and CPSNR, improve image edge blurring, and this method could well meet the requirements of endoscope detection tech- nology in terms of the image effect.