Integrated image processor
First Claim
Patent Images
1. An integrated image processor implemented on a substrate comprising:
- an input interface configured to receive pixel data from one or more images; and
a pixel handling processor disposed on the substrate configured to convert the pixel data from the input interface into depth pixel data; and
a foreground detector processor disposed on the substrate configured to classify pixels from the pixel handling processor as background or not background, wherein the output of the foreground detector processor comprises one bit per pixel indicating whether the pixel is classified as background or not background, and wherein a background pixel comprises a pixel that is determined to be relatively stable using a model of the pixel'"'"'s behavior over time, and wherein the pixel handling processor and the foreground detector are arranged in a low latency pipelined architecture such that each are operating on pixel data from the one or more input images at the same time, and wherein the foreground detector processor includes storage for the model of each pixel'"'"'s behavior over time, wherein the pixel handling processor only requires a subset of its input image to produce an output depth pixel data, and wherein the foreground detector only requires a subset of an input range image to classify a pixel as foreground or background.
3 Assignments
0 Petitions
Accused Products
Abstract
An integrated image processor implemented on a substrate is disclosed. An input interface is configured to receive pixel data from two or more images. A pixel handling processor disposed on the substrate is configured to convert the pixel data into depth and intensity pixel data. In some embodiments, a foreground detector processor disposed on the substrate is configured to classify pixels as background or not background. In some embodiments, a projection generator disposed on the substrate is configured to generate a projection in space of the depth and intensity pixel data.
-
Citations
27 Claims
-
1. An integrated image processor implemented on a substrate comprising:
-
an input interface configured to receive pixel data from one or more images; and a pixel handling processor disposed on the substrate configured to convert the pixel data from the input interface into depth pixel data; and a foreground detector processor disposed on the substrate configured to classify pixels from the pixel handling processor as background or not background, wherein the output of the foreground detector processor comprises one bit per pixel indicating whether the pixel is classified as background or not background, and wherein a background pixel comprises a pixel that is determined to be relatively stable using a model of the pixel'"'"'s behavior over time, and wherein the pixel handling processor and the foreground detector are arranged in a low latency pipelined architecture such that each are operating on pixel data from the one or more input images at the same time, and wherein the foreground detector processor includes storage for the model of each pixel'"'"'s behavior over time, wherein the pixel handling processor only requires a subset of its input image to produce an output depth pixel data, and wherein the foreground detector only requires a subset of an input range image to classify a pixel as foreground or background. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 25)
-
-
9. An integrated image processor implemented on a substrate comprising:
-
an input interface configured to receive pixel data from one or more images; a pixel handling processor disposed on the substrate configured to convert the pixel data from the input interface into depth pixel data; and a projection generator disposed on the substrate configured to generate a projection in space of the depth pixel data received from the pixel handling processor, wherein the projection applies a 3D rotation and translation to the depth data and maps the transformed depth pixel data into a plurality of cells, and wherein one or more statistics for each cell of the plurality of cells are computed based at least in part on the depth pixel data mapped into each cell, and wherein the pixel handling processor and the projection generator are arranged in a low latency pipelined architecture such that each are operating on pixel data from the one or more input images at the same time, and wherein the pixel handling processor only requires a subset of its input data to produce an output depth pixel data, and wherein the projection generator only requires a subset of its input data to update a first cell in an output array of the projection generator, and wherein the projection generator processor includes storage for the one or more statistics for each cell of the plurality of cells. - View Dependent Claims (10, 11, 12, 13, 14, 15, 26)
-
-
16. An integrated image processor implemented on a substrate comprising:
-
an input interface configured to receive pixel data from one or more images; and a pixel handling processor disposed on the substrate configured to convert the pixel data from the input interface into depth pixel data; and a foreground detector processor disposed on the substrate configured to classify depth pixels from the pixel handling processor as background or not background, wherein a background pixel comprises a pixel that is determined to be relatively stable using a model of the pixel'"'"'s behavior over time; a projection generator disposed on the substrate configured to generate a projection in space of the depth pixel data determined by the pixel handling processor, wherein the projection applies a 3D rotation and translation to the depth pixel data and maps the transformed depth pixel data into a plurality of cells, and wherein one or more statistics for each cell of the plurality of cells are computed based at least in part on the depth pixel data mapped into each cell, and wherein the pixel handling processor, the foreground detector, and the projection generator are arranged in a low latency pipelined architecture such that each are operating on pixel data from the one or more input images at the same time, and wherein the projection generator also receives the output of the foreground detector and determines, based at least in part on the one bit per pixel output from the foreground detector, whether a depth pixel is not background, and in the event that the depth pixel is not background, the projection generator processes the depth pixel, and in the event that the depth pixel is background, the projection generator does not process the depth pixel, and wherein the pixel handling processor only requires a subset of its input data to produce an output depth pixel, and wherein the projection generator only requires a subset of its input data to update a first cell in an output array of the projection generator, and wherein the projection generator processor includes storage for the one or more statistics for each cell of the plurality of cells, and wherein the foreground detector processor includes storage for the model of the pixel'"'"'s behavior over time, and wherein the foreground detector only requires a subset of its input image data to produce an output pixel; a segmentation processor disposed on the substrate configured to compute from the output of the projection generator a list of zero or more individual object descriptions, wherein one description of the list comprises a 3D location and a 3D physical extent; and an application processor disposed on the substrate configured to receive the list of zero or more individual object descriptions output by the segmentation processor. - View Dependent Claims (17, 18, 19, 20, 21, 22, 23, 24, 27)
-
Specification