Method and device for temporal filtering of disparity maps
View Patent ↗The invention consists in a method of and a device for temporal filtering of disparity maps of n different frames, each map indicating the disparity of pixels of 3D images comprising the steps of marking the stationary pixels and the non stationary pixels and implementing temporal filter for stationary pixels, detecting limitations for temporal filtering corresponding to a variation of moving pixels above a determined threshold value and automatically de-activating the temporal filters as response to the detected limitations for temporal filtering.
1. A method of temporal filtering of n different successive disparity maps obtained from n different successive frames, n being an integer greater or equal to 2, each frame having a left image and a right image, each disparity map being associated with a frame, and indicating the disparity of pixels of images of said frame, said method comprising:
for a current disparity map, marking the stationary pixels and the non-stationary pixels according to a comparison of pixel disparity values and a first threshold, and implementing temporal filter for stationary pixels; and
de-activating the temporal filters for at least said current disparity map when a variation of the number of non-stationary pixels from said current disparity map and at least one previous disparity map from said n different successive disparity maps, is above a second threshold value.
2. The method of temporal filtering according to claim 1 , further comprising reactivating the temporal filters when said variation is below said second threshold value for at least during the processing of a following disparity map.
3. The method of temporal filtering according to claim 1 , wherein said de-activating the temporal filters for at least said current disparity map is done when variations of the number of non-stationary pixels from said current disparity map and at least each of four previous disparity maps from said n different successive disparity maps, is above a third threshold value.
4. The method of temporal filtering according to claim 3 , wherein said variations are measured in percentages, and said third threshold value is 15%.
5. The method of temporal filtering according to claim 3 , wherein said variations are measured in percentages, and said third threshold value is 25%.