US 7,161,614 B1Granted

Device and method for converting two-dimensional video to three-dimensional video

AssigneeSanyo Electric Co., Ltd.

InventorsSyugo Yamashita, Haruhiko Murata, Toshiya Iinuma, Mitsuo Nakashima, Takayuki Mori

Life of this patent

Estimating the term…

Loading patent…

Abstract

There is provided parallax correction means for correcting a parallax for each area calculated by parallax calculation means in accordance with the magnitude of a motion vector for the area detected by motion vector detection means in order to prevent the three-dimensional effect of a conversion video from greatly differing depending on an input video when the MTD method and the CID method are simultaneously used. When a depth estimate is converted into a parallax, the depth estimate is subjected to distance scale conversion in order to suppress the distortion of the conversion image, to find a tentative target phase for each parallax calculation area, and a dynamic range in which a phase difference between the parallax calculation areas is within a distortion allowable range is searched for and is subjected to distance scale conversion, to find a tentative target phase, which operations are repeated.

Claims8

Dependent claims

A device for converting a two-dimensional video into a three-dimensional video, characterized by comprising:

a field memory for storing for each field a two-dimensional video signal inputted from a video signal source;
motion vector detection means for detecting for each area of an input video a motion vector corresponding to movement between fields of the inputted video signal;
readout means for reading out, out of the video signals stored in the field memory, the video signal delayed from the inputted video signal by a delay found from the motion vector for each area detected by the motion vector detection means from the field memory;
switching means for outputting one of the inputted video signal and the video signal read out of the field memory and the other video signal, respectively, as a left eye video signal and a right eye video signal on the basis of the direction of a horizontal component of the motion vector for each area detected by the motion vector detection means;
feature extraction means for extracting for each area of the input video a video feature from the inputted video signal;
parallax calculation means for calculating, on the basis of the image feature for each area of the input video extracted by the feature extraction means, a depth for the area and calculating a parallax for the area from the calculated depth for the area;
parallax correction means for correcting the parallax for each area calculated by the parallax calculation means depending on the magnitude of the motion vector for the area detected by the motion vector detection means;
and
phase control means for correcting, on the basis of the parallax for each area corrected by the parallax correction means, phases for the area of the right eye video and the left eye video outputted by the switching means, and outputting the videos as a stereoscopic video signal.

6 more elements

A method of converting a two-dimensional video into a three-dimensional video, characterized by comprising:

a first step of storing for each field a two-dimensional video signal inputted from a video signal source in a field memory;
a second step of detecting for each area of an input video a motion vector corresponding to movement between fields of the inputted video signal;
a third step of reading out, out of the video signals stored in the field memory, a video signal delayed from the inputted video signal by a delay found from the motion vector for each area detected at the second step from the field memory;
a fourth step of outputting one of the inputted video signal and the video signal read out of the field memory and the other video signal, respectively, as a left eye video signal and a right eye video signal on the basis of the direction of a horizontal component of the motion vector for each area detected at the second step;
a fifth step of extracting for each area of the input video a video feature from the inputted video signal;
a sixth step of calculating, on the basis of the image feature for each area of the input video extracted at the fifth step, a depth for the area and calculating a parallax for the area from the calculated depth for the area;
a seventh step of correcting the parallax for each area calculated at the sixth step depending on the magnitude of the motion vector for the area detected at the second step;
and
an eighth step of correcting, on the basis of the parallax for each area corrected at the seventh step, phases for the area of the right eye video and the left eye video outputted at the fourth step, and outputting the videos as a stereoscopic video signal.

6 more elements

A method of converting a two-dimensional video into a three-dimensional video, characterized by comprising:

a first step of extracting an image feature related to the long or short distance of a video from each of a plurality of parallax calculation areas set within one screen on the basis of a two-dimensional video signal, and producing a depth estimate for the parallax calculation area on the basis of the extracted image feature;
a second step of subjecting each of the depth estimates to distance scale conversion using a dynamic range defined by a predetermined maximum projection amount and a predetermined maximum recession amount, to find a tentative target phase for each of the parallax calculation areas;
a third step of finding the maximum value of a phase difference between the adjacent parallax calculation areas on the basis of the tentative target phase for each of the parallax calculation areas;
a fourth step of judging whether or not the maximum value of the phase difference between the adjacent parallax calculation areas is within a predetermined distortion allowable range;
and
a fifth step of searching, when the maximum value of the phase difference between the adjacent parallax calculation areas is outside the predetermined distortion allowable range, for such a dynamic range that the phase difference between said parallax calculation areas is within the distortion allowable range, subjecting each of the depth estimates to distance scale conversion using the dynamic range searched for, and finding a tentative target phase for each of the parallax calculation areas, to proceed to the third step.

3 more elements

Dependencies are inferred from the claim text.