US 10,096,122 B1Granted

Segmentation of object image data from background image data

AssigneeAmazon Technologies, Inc.

InventorsAmit Kumar Agrawal, Abhishek Singh, Visesh Uday Kumar Chari, Lelin Zhang, Qiang Liu, Rohith Mysore Vijaya Kumar, David Ting-Yu Wu

Life of this patent

Estimating the term…

patentgrep

Loading patent…

Abstract

Devices and techniques are generally described for segmentation of object image data from background image data. In some examples, the segmentation of object image data may comprise capturing image data comprising color data and depth data. In some examples, the segmentation of object image data may further include separating the depth data into a plurality of clusters of image data, wherein each cluster is associated with a respective range of depth values. In various examples, the segmentation of object image data may comprise selecting a main cluster of image data as corresponding to an object of interest in the image data. In various other examples, the segmentation of object image data may comprise identifying pixels of the main cluster that correspond to the object of interest.

Claims20

Dependent claims

A method of segmenting a clothing item of interest in an image from other objects in the image, the method comprising:

receiving image data captured by an image capture device, wherein the image data comprises:
(1) color image data representing a scene including a clothing item, and (2) a depth map representing the scene, the depth map comprising depth values indicating a distance of objects in the scene from the image capture device;
separating the depth map into a plurality of clusters of depth map pixels, wherein for each cluster of the plurality of clusters of depth map pixels, each depth map pixel of the cluster of depth map pixels has a respective depth value;
identifying a set of candidate clusters from the plurality of clusters, each of the candidate clusters having a size between a lower threshold size and an upper threshold size, wherein the lower threshold size and the upper threshold size define an expected size range of the clothing item;
identifying a main cluster from the set of candidate clusters, wherein an average depth value of pixels in the main cluster indicates that the main cluster is closest to the image capture device from among the set of candidate clusters;
determining a depth value for each pixel of the main cluster;
encoding a first set of pixels of the main cluster having a respective depth value above a threshold depth value with background indicator data, wherein the threshold depth value corresponds to a position between the clothing item and a user holding the clothing item;
encoding a second set of pixels of the main cluster having a respective depth value below the threshold depth value with foreground indicator data;
generating a cropped image including only the second set of pixels;
and
rendering the cropped image.

8 more elements

An image capture device comprising:

a digital camera module effective to capture color data;
a depth sensor effective to capture depth values;
at least one processor in communication with the digital camera module and the depth sensor, the at least one processor effective to execute instructions that when executed by the at least one processor cause the image capture device to perform a method comprising:
separating the depth values into a plurality of clusters, wherein each cluster includes a portion of the depth values;
selecting a first cluster of the plurality of clusters, the first cluster corresponding to a representation of an object of interest;
determining a depth value of each pixel in the first cluster;
determining a threshold depth value based at least in part on the depth value of each pixel in the first cluster;
encoding a first plurality of pixels in the first cluster with foreground indicator data, wherein each of the first plurality of pixels comprises a depth value less than the threshold depth value;
encoding a second plurality of pixels in the first cluster with background indicator data, wherein each of the second plurality of pixels comprises a depth value greater than the threshold depth value;
and
generating image data representing the object of interest and including the first plurality of pixels.

8 more elements

Dependencies are inferred from the claim text.