US 11,935,257 B2
Adding an adaptive offset term using convolution techniques to a local adaptive binarization expression
Liangliang Wang, Shanghai (CN); Wenhai Gao, Shanghai (CN); and Bo Yu, Shanghai (CN)
Assigned to Ambarella International LP, Santa Clara, CA (US)
Filed by Ambarella International LP, Santa Clara, CA (US)
Filed on Aug. 26, 2021, as Appl. No. 17/412,715.
Claims priority of application No. 202110943775.2 (CN), filed on Aug. 16, 2021.
Prior Publication US 2023/0052553 A1, Feb. 16, 2023
Int. Cl. G06T 7/521 (2017.01); G06F 18/22 (2023.01); G06T 7/00 (2017.01); G06T 7/194 (2017.01); G06T 7/514 (2017.01)
CPC G06T 7/521 (2017.01) [G06F 18/22 (2023.01); G06T 7/0002 (2013.01); G06T 7/194 (2017.01); G06T 7/514 (2017.01); G06T 2207/10016 (2013.01); G06T 2207/20012 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30168 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An apparatus comprising:
an interface configured to receive pixel data;
a structured light projector configured to generate a structured light pattern; and
a processor configured to (i) process said pixel data arranged as video frames, (ii) perform operations using a convolutional neural network to determine a binarization result and an offset value and (iii) generate a disparity map and a depth map in response to (a) said video frames, (b) said structured light pattern, (c) said binarization result, (d) said offset value and (e) a removal of error points, wherein said convolutional neural network:
(A) performs a partial block summation and an average on said video frames to generate a convolution result,
(B) compares said convolution result to an ideal speckle value to determine said offset value,
(C) generates an adaptive result in response to performing a convolution operation to add said offset value to said video frames,
(D) compares said video frames to said adaptive result to generate said binarization result for said video frames, and
(E) removes said error points from said binarization result.