Effects of image positions on temperature reconstruction using light field camera

Abstract

The light field camera has emerged as an optical flame temperature detection device owing to the refocusing feature. Combining this feature with the deconvolution algorithm, a three-dimensional flame temperature distribution can be reconstructed. However, the accuracy of reconstructing the flame temperature is limited by the actual image position. Based on the established flame and light field camera model, the proposed Lucy-Richardson and Nearest Neighbor Filtering joint deconvolution algorithm is applied to sectioned temperature reconstruction. Thus, the effects of image position using light field camera on temperature reconstruction are verified. The relative error of flame temperature reconstruction is less than 10%, if the flame is imaged in the middle of the light field image. However, if the flame is imaged at the image edge, the edge distortion and defocused state of the camera increase the relative error to less than 11% and 14%, respectively. The effects of image positions are not too significant for temperature reconstruction using light field camera.