In this paper we focus on the problem of shadow removal from a single image.
In earlier work, Finlayson et al.[5][6] remove shadows by zeroing shadow edges
in the gradient domain and then integrating to obtain a shadow free image.
They achieve good results with high quality images, however the integration
often introduces changes in color balance, global smoothness and loss of textural
The advantage of superpixels is analyzed and
shown in applications, such as object recognition [35] and segmen-
tation [36]. Fig. 1 shows superpixel segmentation, in which the
image is divided into superpixels and each superpixel shows the
same visual appearance, which can cause substantial speed-up of
subsequent processing. Therefore, the careful choice of the super-
pixel method and its parameters for the particular application are
crucial. We use TurboPixels [37] to extract superpixels from an
image, in which one superpixel is roughly uniform in texture and
gray, so that the boundaries of regions are preserved. In order to
encode gray, texture and spatial information into superpixels, we
describe each superpixel j by a 7-dimensional wavelet feature vec-
tor Fj =(f1, f2, ... , f7), in which Fj is the average wavelet value of all
In this paper we focus on the problem of shadow removal from a single image.In earlier work, Finlayson et al.[5][6] remove shadows by zeroing shadow edgesin the gradient domain and then integrating to obtain a shadow free image.They achieve good results with high quality images, however the integrationoften introduces changes in color balance, global smoothness and loss of texturalThe advantage of superpixels is analyzed andshown in applications, such as object recognition [35] and segmen-tation [36]. Fig. 1 shows superpixel segmentation, in which theimage is divided into superpixels and each superpixel shows thesame visual appearance, which can cause substantial speed-up ofsubsequent processing. Therefore, the careful choice of the super-pixel method and its parameters for the particular application arecrucial. We use TurboPixels [37] to extract superpixels from animage, in which one superpixel is roughly uniform in texture andgray, so that the boundaries of regions are preserved. In order toencode gray, texture and spatial information into superpixels, wedescribe each superpixel j by a 7-dimensional wavelet feature vec-tor Fj =(f1, f2, ... , f7), in which Fj is the average wavelet value of all
การแปล กรุณารอสักครู่..
