Improving color constancy by discounting the variation of camera spectral sensitivity

Abstract

It is an ill-posed problem to recover the true scene colors from a color biased image by discounting the effects of scene illuminant and camera spectral sensitivity (CSS) at the same time. Most color constancy (CC) models have been designed to first estimate the illuminant color, which is then removed from the color biased image to obtain an image taken under white light, without the explicit consideration of CSS effect on CC. This paper first studies the CSS effect on illuminant estimation arising in the inter-dataset-based CC (inter-CC), i.e., training a CC model on one dataset and then testing on another dataset captured by a distinct CSS. We show the clear degradation of existing CC models for inter-CC application. Then a simple way is proposed to overcome such degradation by first learning quickly a transform matrix between the two distinct CSSs (CSS-1 and CSS-2). The learned matrix is then used to convert the data (including the illuminant ground truth and the color-biased images) rendered under CSS-1 into CSS-2, and then train and apply the CC model on the color-biased images under CSS-2 without the need of burdensome acquiring of the training set under CSS-2. Extensive experiments on synthetic and real images show that our method can clearly improve the inter-CC performance for traditional CC algorithms. We suggest that, by taking the CSS effect into account, it is more likely to obtain the truly color constant images invariant to the changes of both illuminant and camera sensors.

Revisiting Gray Pixel for Statistical Illumination Estimation

“…In the literature, the validation of color constancy methods is customarily performed using k-fold cross-validation on the same dataset. As a result, this validation process favors learning-based methods and fails to assess their performance for color correction in images from an unknown camera (Gao et al, 2017).…”

“…We ar-gue that learning-based methods depend on the assumption that the statistical distribution of the illumination in both the training and testing images is similar. In other words, learning-based methods assume that imaging and illumination conditions of a given image can be inferred from previous training examples, thus becoming heavily dependent on the training data (Gao et al, 2017).…”

Unsupervised Learning for Color Constancy

“…Use angular distance next step is to extend it so that it can train on images taken with one sensor and be used on images taken with another sensor. A solution to this problem has already been proposed [43], but it requires calibrated images and reflectance spectras to learn a 3 × 3 sensor transformation matrix. Here, however, the goal is to use neither calibrated images nor any reflectance spectras in order for the method to be fully unsupervised.…”

“…Nevertheless, since all well-known learning-based methods are supervised, a major obstacle for their application is that for a given sensor, despite proposed workarounds [43], supervised learning-based methods have to be trained on calibrated images taken by preferably the same sensor [44]. To calibrate arXiv:1712.00436v4 [cs.CV] 19 Mar 2019 the images, a calibration object has to be placed in the scenes of these images and later segmented to extract the groundtruth illumination.…”

Color Constancy Convolutional Autoencoder

“…As in INTEL-TUT2 dataset different camera models are used, the variation of camera spectral sensitivity needs to be discounted. For this purpose, we utilize Color Conversion Matrix (CCM) based preprocessing [28] to learn 3*3 CCM matrices for each camera pair.…”