Considering the camera imaging pipeline for color image processing: Color stabilization, blind gamma estimation, and…
Place: Large lecture room.
Affiliation: Universitat Pompeu Fabra, Barcelona, Spain.
Considering the camera imaging pipeline for color image processing: Color stabilization, blind gamma estimation, and color characterization
Two main topics will be presented in this seminar.
In the first topic, we will present the works of [1] and [2]. In particular, we will focus our attention in two crucial but often overlooked observations from the camera imaging pipeline: firstly, that the core of the color correction chain in a digital camera is simply a multiplication by a 3x3 matrix; secondly, that to color-match a source image to a reference image we don’t need to compute their two color correction matrices, it is enough to compute the operation that transforms one matrix into the other. From these observations, we will propose a method for color stabilization of shots of the same scene [1], taken under the same illumination, where one image is chosen as a reference and one or several other images are modified so that their colors match those of the reference. Our approach only requires a set of pixel correspondences, it does not need any information about the cameras used, nor models or specifications or parameter values. There is a wide range of applications of our technique, both for amateur and professional photography and video: color matching for multi-camera TV broadcasts, color matching for 3D cinema, color stabilization for amateur video, etc. Later, we will present an extension of the aforementioned method in order to perform gamma correction from and for multiple images [2]. The expression “from and for” is used as we need more than a single image to obtain the gamma estimation but, at the same time, we recover as many estimates as images are used.
The second topic of the seminar regards the work of [3]. In this case, we will show how image-based perceptual metrics can be used to characterize any color camera. To this end, we will make use of the Spherical Sampling technique [4], where color sensors are related to points in a sphere, therefore being easier to sample. Our approach outperforms previous approaches by an amount varying between 3% and 13% depending on the measure considered.
[1] Color Stabilization Along Time and Across Shots of the Same Scene, for One or Several Cameras of Unknown Specifications, Javier Vazquez-Corral, Marcelo Bertalmío IEEE Transactions on Image Processing, 23(10):4564-75, 2014.
[2] Simulateneous blind gamma estimation, Javier Vazquez-Corral, Marcelo Bertalmío, IEEE Signal Processing Letters, accepted.
[3] Perceptual Color Characterization of Cameras, Javier Vazquez-Corral, David Connah, Marcelo Bertalmío, Sensors 2014, 14(12), 23205-23229
[4] Spectral sharpening by spherical sampling, Graham Finlayson, Javier Vazquez-Corral, Sabine Süsstrunk, Maria Vanrell, Journal of the Optical Society of America -A (JOSA A) 29 (7), 1199-1210, 2012
SHARE ON
