Method and system for zone-wise adaptive illumination of objects

Patent metadata

Field	Value
Publication number	US-12363443-B2
Application number	US-202218054832-A
Country	US
Kind code	B2
Filing date	Nov 11, 2022
Priority date	Nov 29, 2021
Publication date	Jul 15, 2025
Grant date	Jul 15, 2025

Abstract

Lighting conditions affect quality of images being captured. In traditional systems, illuminators (light sources) are used to illuminate the objects without considering illumination levels at different zones/sides of an object being photographed. As a result, all illuminators may run at maximum capacity, resulting in wastage of power and compromising efficiency of the system. The disclosure herein generally relates to illumination of objects, and, more particularly, to a method and system for adaptive illumination of objects. The system determines illumination at different zones of the object, and further identifies zones that are not illuminated properly in comparison with a threshold of illumination. Further the system controls intensity of only the illuminators which are responsible for illumination of the zones in which measured illumination is below a threshold of illumination, and increases the intensity by a value determined based on difference between measured illumination and the threshold of illumination, for each zone.

First claim

What is claimed is:

1. A processor implemented method for adaptive illumination of an object, comprising:
obtaining at least one image of an object, via one or more hardware processors, wherein the at least one image is captured when the object is illuminated by a plurality of Infrared (IR) illuminators from an IR illuminator array;
pre-processing the at least one image, via the one or more hardware processors, wherein the pre-processing comprises of a) adjusting contrast, and b) performing normalization, of one or more frames captured in the at least one image;
performing a semantic segmentation of pre-processed at least one image, via the one or more hardware processors, to extract foreground (FG) data and background (BG) data of the at least one image;
generating a binary semantic mask of the at least one image, via the one or more hardware processors, wherein the binary semantic mask comprises information on the extracted BG data and the FG data of the at least one image, wherein at least one Region of Interest (RoI) with a FG is identified, based on the data in the binary semantic mask of the at least one image;
performing a full scale segmentation of all objects in each of the ROIs in the FG of the at least one image, via the one or more hardware processors, wherein when the object is an ore material and application is ore fragmentation analysis, then the full scale segmentation identifies impurities as rock particles, and further to address overlapping un-segmented regions, the steps comprise:
determining bounding boxes, using an additional regional proposal network, and segment out overlapping regions by eroding corresponding boundaries, wherein the determined overlapping regions is annotated with boulders, and corresponding parameters pertains to a boulder count, a mean particle size are evaluated;

constructing an Edge based Structural Similarity Index Metric (ESSIM) matrix from the binary semantic mask, via the one or more hardware processors, wherein the ESSIM indicates a measured illuminance of each of a plurality of zones in the at least one Region of Interest (RoI) of the object, in terms of a) luminance (L), b) contrast (I), and c) one or more edge comparison functions (E), wherein the luminance (L) is estimated as: L(x, y)=(2μ x μ y +S 1 )/(μ x 2 +μ y 2 +S 1 ), wherein the contrast (I) is estimated as: C(x, y)=(2σ x σ y +S 2 )/(σ x 2 +σ y 2 +S 2 ), wherein the one or more edge comparison functions (E) is estimated as: E(x, y)=(ρ xy +S 3 )/(ρ x ρ y +S 3 ), wherein S 1 , S 2 , and S 3 are constants, ρ x is a standard deviation of edge direction vector, ρ xy is a co-variance of direction vectors corresponding to x and y respectively;
determining, via the one or more hardware processors, whether the measured illuminance of each of the plurality of zones of the object at least matches a threshold of illuminance; and
varying, via the one or more hardware processors, intensity of the plurality of illuminators, to improve the illuminance of each of the plurality of zones for which the measured illuminance is below the threshold of illuminance, to at least match the threshold of illuminance,
wherein varying the intensity of the plurality of illuminators comprises:
identifying the plurality of illuminators corresponding to each of the plurality of zones for which the measured illuminance is below the threshold of illuminance, based on a mapping of each of the plurality of zones with the corresponding one or more illuminators;
determining extent to which the intensity of each of the identified one or more illuminators is required to be varied to improve the illumination in the corresponding zone of at least one image to at least match the threshold of illuminance, wherein the step of determining whether or not to control intensity of each of the IR emitters, based on a difference in the measured illumination and the threshold of illuminance, serves as a feedback mechanism allowing control of the intensity of the IR emitters only when the measured illuminance of at least one of the zone does not match the threshold of illuminance;
generating a control signal to vary the intensity of the one or more of the plurality of illuminators, based on the determined extent to which the intensity of each of the identified one or more illuminators is required to be varied, wherein zone-wise estimated control signals are used to vary and adapt the intensity of the IR illuminators, and
varying the intensity of each of the identified one or more illuminators, using the generated control signal.

2. The method of claim 1 , wherein the one or more edge comparison functions indicate whether a plurality of edges of the binary semantic mask match a threshold of mask.

3. The method of claim 1 , wherein the ore fragmentation analysis of the ore material is performed after varying the intensity of the one or more illuminators.

4. A system for adaptive illumination of an object, comprising:
one or more hardware processors;
a communication interface; and
a memory storing a plurality of instructions, wherein the plurality of instructions when executed, cause the one or more hardware processors to:
obtain at least one image of an object, wherein the at least one image is captured when the object is illuminated by a plurality of Infrared (IR) illuminators from an IR illuminator array;
pre-process the at least one image, wherein pre-processing the at least one image comprises of a) adjusting contrast, and b) performing normalization, of one or more frames captured in the at least one image;
perform a semantic segmentation of pre-processed at least one image to extract foreground (FG) data and background (BG) data of the at least one image;
generate a binary semantic mask of the at least one image, wherein the binary semantic mask comprises information on the extracted BG data and the FG data of the at least one image, wherein at least one Region of Interest (RoI) with a FG is identified, based on the data in the binary semantic mask of the at least one image;
perform a full scale segmentation of all objects in each of the ROIs in the FG of the at least one image, wherein when the object is an ore material and application is ore fragmentation analysis, then the full scale segmentation identifies impurities as rock particles, and further to address overlapping un-segmented regions, the steps comprise:
determine bounding boxes, using an additional regional proposal network, and segment out overlapping regions by eroding corresponding boundaries, wherein the determined overlapping regions is annotated with boulders, and corresponding parameters pertains to a boulder count, a mean particle size are evaluated;

construct an Edge based Structural Similarity Index Metric (ESSIM) matrix from the binary semantic mask, wherein the ESSIM indicates a measured illuminance of each of a plurality of zones in the at least one Region of Interest (RoI) of the object, in terms of a) luminance (L), b) contrast (I), and c) one or more edge comparison functions (E), wherein the luminance (L) is estimated as: L(x, y)=(2μ x μ y +S 1 )/(μ x 2 +μ y 2 +S 1 ), wherein the contrast (I) is estimated as: C(x, y)=(2σ x σ y +S 2 )/(σ x 2 +σ y 2 +S 2 ), wherein the one or more edge comparison functions (E) is estimated as: E(x, y)=(ρ xy +S 3 )/(ρ x ρ y +S 3 ), wherein S 1 , S 2 , and S 3 are constants, ρ x is a standard deviation of edge direction vector, ρ xy is a co-variance of direction vectors corresponding to x and y respectively;
determine whether the measured illuminance of each of the plurality of zones of the object at least matches a threshold of illuminance; and
vary intensity of the plurality of illuminators, to improve the illuminance of each of the plurality of zones for which the measured illuminance is below the threshold, to at least match the

Assignees

• Tata Consultancy Services Ltd

Inventors

• Swain Amit
• Ray Arindam
• Chakravarty Tapas
• Bhaumik Chirabrata
• Rani Smriti

Classifications

• H04N23/71

  Circuitry for evaluating the brightness variation · CPC title

• G06V10/443

  by matching or filtering · CPC title

• G06T2207/20112

  Image segmentation details · CPC title

• G06T2207/10152

  Varying illumination · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title