Back to results
US 12,361,707 B1Granted
System and method for discriminating and demarcating targets of interest in a physical scene
AssigneeRenesas Electronics America Inc.
InventorJeffrey Mark Sieracki
Life of this patent
Estimating the term…
Abstract
Captured samples of a physical structure or other scene are mapped to a predetermined multi-dimensional coordinate space, and spatially-adjacent samples are organized into array cells representing subspaces thereof. Each cell is classified according to predetermined target-identifying criteria for the samples of the cell. A cluster of spatially-contiguous cells of common classification, peripherally bounded by cells of different classification, is constructed, and a boundary demarcation is defined from the peripheral contour of the cluster. The boundary demarcation is overlaid upon a visual display of the physical scene, thereby visually demarcating the boundaries of a detected target of interest.
Claims19
A system for detecting and demarcating a target of interest represented within a physical scene, the system comprising:
- an array generating portion executing on a processor to construct a multi-dimensional spatial array of cells defined according to a predetermined multi-dimensional coordinate space, each cell of the spatial array containing a plurality of spatially-adjacent samples respectively mapped to corresponding spatially-adjacent coordinates within the predetermined multi-dimensional coordinate space, each sample of the plurality of spatially-adjacent samples captured by a transducing device, each cell of the spatial array thereby spanning a subspace within the predetermined multi-dimensional coordinate space;
- a target discriminating portion executing on a processor to:
- classify at least one cell of the spatial array of cells according to predetermined target-identifying criteria for the samples of the at least one cell, the plurality of spatially-adjacent samples of each cell of the at least one classified cell being thereby respectively assigned a common cell-wide classification, each sample of the plurality of spatially-adjacent samples of the at least one cell corresponding to a different respective portion of the physical scene, and
- construct at least one cluster of a plurality of spatially-contiguous cells of common classification according to the classified at least one cell, the at least one cluster being peripherally bounded in the predetermined multi-dimensional coordinate space by cells of different classification, the at least one cluster at least partially defining the target of interest in peripheral contour within the physical scene, wherein the target discriminating portion classifies the at least one cell of the spatial array by:
- extracting a plurality of data vectors from differing respective sets of samples of the plurality of spatially-adjacent samples of said at least one cell, wherein each data vector of said plurality of data vectors corresponding to said at least one cell is extracted independently from one another,
- classifying said each independently extracted data vector of said plurality of data vectors independently from one another according to the predetermined target-identifying criteria to produce a respective classification score for said each independently extracted data vector of said plurality of data vectors corresponding to said at least one cell,
- combining, by voting, said each respective classification score for said each independently extracted and classified data vector of said plurality of data vectors corresponding to said at least one cell to result in a combined score for said at least one cell, and
- classifying said at least one cell according to said combined score for said at least one cell;
- and
- an interface portion configured to generate a composite image having at least one visual demarcation overlaying a visual display of the physical scene, the at least one visual demarcation being defined with respect to the predetermined multi-dimensional coordinate space according to the peripheral contour of the at least one cluster and being overlaid upon corresponding coordinates of the visual display of the physical scene.
7 more elements
A method for detecting and visually demarcating a target of interest represented within a physical scene, the method comprising:
- capturing a plurality of samples of the physical scene;
- mapping each sample of the plurality of captured samples to corresponding coordinates within a predetermined multi-dimensional coordinate space;
- constructing a multi-dimensional spatial array of cells defined to span a subspace within the predetermined multi-dimensional coordinate space, each cell of the spatial array containing a plurality of spatially-adjacent samples of the plurality of captured samples respectively mapped to corresponding spatially-adjacent coordinates within the subspace;
- classifying at least one cell of the spatial array of cells according to predetermined target-identifying criteria for the samples of the at least one cell, the plurality of spatially-adjacent samples of each cell of the at least one classified cell being thereby respectively assigned a common cell-wide classification, each sample of the plurality of spatially-adjacent samples of the at least one cell corresponding to a different respective portion of the physical scene, wherein the classification of the at least one cell of the spatial array includes:
- extracting a plurality of data vectors from differing respective sets of samples of the plurality of spatially-adjacent samples of said at least one cell, wherein each data vector of said plurality of data vectors corresponding to said at least one cell is extracted independently from one another,
- classifying said each independently extracted data vector of said plurality of data vectors independently from one another according to the predetermined target-identifying criteria to produce a respective classification score for said each independently extracted data vector of said plurality of data vectors corresponding to said at least one cell,
- combining, by voting, said each respective classification score for said each independently extracted and classified data vector of said plurality of data vectors corresponding to said at least one cell to result in a combined score for said at least one cell, and
- classifying said at least one cell according to said combined score for said at least one cell;
- constructing at least one cluster of a plurality of spatially-contiguous cells of common classification according to the classified at least one cell, the at least one cluster being peripherally bounded in the predetermined multi-dimensional coordinate space by cells of different classification, the at least one cluster at least partially defining the target of interest in peripheral contour within the physical scene;
- and
- generating a composite image having at least one visual demarcation overlaying a visual display of the physical scene, the at least one visual demarcation being defined with respect to the predetermined multi-dimensional coordinate space according to the peripheral contour of the at least one cluster and being overlaid upon corresponding coordinates of the visual display of the physical scene.
8 more elements
Dependencies are inferred from the claim text.