Granted Patent B2

US 12,482,079 · App. 17/917,303 · Granted Nov 25, 2025

Inventors: Mahendra Kumar Angamuthu Ganesan (Chennai, IN); Prabu Palaniappan (Namakkal, IN); Arun Anandan (Chennai, IN); Swaran Thekkevavanoor (Chennai, IN); Abdul A. Mahaboob Basha (Chennai, IN)

Assignee: Caterpillar Inc.

G06T5/80G06T3/40G06T5/20G06V10/60G06V10/751G06T2207/20028

Abstract

Claims (81)

1 . A method for removing haziness in a digital image, the method comprising:

receiving, by a processing unit, an input digital image from an image capturing device, the input digital image depicting at least some haze content;

downscaling, by the processing unit, the input digital image to obtain a low-resolution image, wherein the low-resolution image comprises pixels having intensity values associated with multiple color channels;

determining, by the processing unit, a dark channel image corresponding to the low-resolution image by:

identifying minimum intensity dark channels associated with respective local patches of the pixels of the low-resolution image, wherein:

a minimum intensity dark channel associated with a particular local patch is identified as one color channel of the multiple color channels that is associated with lowest intensity values within the particular local patch; and

generating the dark channel image using the minimum intensity dark channels associated with the respective local patches;

determining, by the processing unit, an atmospheric light value associated with the low-resolution image;

determining, by the processing unit, a transmission map of the low-resolution image by computing a transmission ratio for each of the respective local patches, based on:

the minimum intensity dark channels associated with the respective local patches in the dark channel image, and

the atmospheric light value;

scaling, by the processing unit, the transmission map to a corresponding magnitude of the input digital image to generate a scaled transmission map;

generating, by the processing unit, a de-hazed output image by applying the scaled transmission map and the atmospheric light value to the input digital image; and

displaying, by the processing unit, the de-hazed output image on a display unit.

2 . The method of claim 1 , wherein receiving the input digital image further comprises:

receiving, by an encoder associated with the image capturing device, a video feed captured by the image capturing device;

encoding, by the encoder, the video feed into a compressed motion-JPEG image format; and

decoding, by a decoder associated with the display unit, the compressed motion-JPEG image format to obtain the input digital image in a YUV image format.

3 . The method of claim 1 , wherein the low-resolution image has a resolution of 320 pixels wide and 180 pixels high.

4 . The method of claim 1 , wherein determining the transmission map further comprises refining, by the processing unit, the transmission map using a bilateral filtering technique.

5 . The method of claim 1 , wherein determining the atmospheric light value of the low-resolution image comprises:

identifying, by the processing unit, a set of brightest pixels in the dark channel image;

identifying, by the processing unit, a set of high intensity pixels of the input digital image that correspond to the set of brightest pixels in the dark channel image; and

determining, by the processing unit, the atmospheric light value of the low-resolution image based on the intensity values of the set of high intensity pixels of the input digital image.

6 . The method of claim 1 further comprising enhancing, by the processing unit, one or more of brightness or contrast characteristics of the de-hazed output image.

7 . A system for removing haziness in a digital image, the system comprising:

an image capturing device associated with a mobile work machine;

a display unit associated with the mobile work machine;

a memory unit; and

a processing unit communicatively coupled to the image capturing device, the display unit, and the memory unit, the processing unit being configured to:

receive an input digital image, from the image capturing device, depicting at least some haze content;

downscale the input digital image to obtain a low-resolution image, wherein the low-resolution image comprises pixels having intensity values associated with multiple color channels;

determine a dark channel image corresponding to the low-resolution image by:

identifying minimum intensity dark channels associated with respective local patches of the pixels of the low-resolution image, wherein:

a minimum intensity dark channel associated with a particular local patch is identified as one color channel of the multiple color channels that is associated with lowest intensity values within the particular local patch; and

generating the dark channel image using the minimum intensity dark channels associated with the respective local patches;

determine an atmospheric light value associated with the low-resolution image;

determine a transmission map of the low-resolution image by computing a transmission ratio for each of the respective local patches, based on:

the minimum intensity dark channels associated with the respective local patches in the dark channel image, and

the atmospheric light value;

scaling, by the processing unit, the transmission map to a corresponding magnitude of the input digital image to generate a scaled transmission map;

generating a de-hazed output image by applying the scaled transmission map and the atmospheric light value to the input digital image; and

display the de-hazed output image on the display unit.

8 . The system of claim 7 , further comprising:

an encoder configured to receive a video feed captured by the image capturing device and encode the video feed into a compressed motion-JPEG image format; and

a decoder configured to decode the compressed motion-JPEG image format to obtain the input digital image in a YUV image format.

9 . The system of claim 7 , wherein the low-resolution image has a resolution of 320 pixels wide and 180 pixels high.

10 . The system of claim 7 , wherein the processing unit is a low power graphics processing unit.

11 . The system of claim 7 , wherein the processing unit is further configured to refine the transmission map by using a bilateral filtering technique.

12 . The system of claim 7 , wherein the processing unit is configured to determine the atmospheric light value of the low-resolution image by:

identifying a set of brightest pixels in the dark channel image;

identifying a set of high intensity pixels of the input digital image that correspond to the set of brightest pixels in the dark channel image; and

determining the atmospheric light value of the low-resolution image based on the intensity values of the set of high intensity pixels of the input digital image.

13 . The system of claim 7 , wherein the processing unit is further configured to enhance one or more of brightness or contrast characteristics of the de-hazed output image.

14 . The method of claim 2 , wherein:

receiving the input digital image further comprises converting, by the processing unit, the input digital image in the YUV image format into an RGB image format, and

the multiple color channels comprise a red channel, a green channel, and a blue channel associated with the RGB image format.

15 . The method of claim 1 , wherein the atmospheric light value is determined based on at least one of:

a constant value provided by a light meter, or

a look-up table based on at least one of a time of day or a location associated with the input digital image.

16 . One or more non-transitory computer-readable media storing computer-executable instructions that, when executed by a processing unit, cause the processing unit to:

receive an input digital image depicting at least some haze content;

downscale the input digital image to obtain a low-resolution image, wherein the low-resolution image comprises pixels having intensity values associated with multiple color channels;

determine a dark channel image corresponding to the low-resolution image by:

identifying minimum intensity dark channels associated with respective local patches of the pixels of the low-resolution image, wherein:

a minimum intensity dark channel associated with a particular local patch is identified as one color channel of the multiple color channels that is associated with lowest intensity values within the particular local patch; and

generating the dark channel image using the minimum intensity dark channels associated with the respective local patches;

determine an atmospheric light value associated with the low-resolution image;

determine a transmission map of the low-resolution image by computing a transmission ratio for each of the respective local patches, based on:

the minimum intensity dark channels associated with the respective local patches in the dark channel image, and

the atmospheric light value;

generate a scaled transmission map by scaling the transmission map to a corresponding magnitude of the input digital image;

generate a de-hazed output image by applying the scaled transmission map and the atmospheric light value to the input digital image; and

cause a display unit to display the de-hazed output image.

17 . The one or more non-transitory computer-readable media of claim 16 , wherein the multiple color channels comprise a red channel, a green channel, and a blue channel associated with a RGB image format.

18 . The one or more non-transitory computer-readable media of claim 17 , wherein the minimum intensity dark channels associated with the respective local patches are identified as one of the red channel, the green channel, or the blue channel that is associated with the lowest intensity values within the respective local patches.

19 . The one or more non-transitory computer-readable media of claim 16 , wherein determining the transmission map further comprises refining, by the processing unit, the transmission map using a bilateral filtering technique.

20 . The one or more non-transitory computer-readable media of claim 16 , wherein determining the atmospheric light value of the low-resolution image comprises:

identifying a set of brightest pixels in the dark channel image;

identifying a set of high intensity pixels of the input digital image that correspond to the set of brightest pixels in the dark channel image; and

determining, by the processing unit, the atmospheric light value of the low-resolution image based on the intensity values of the set of high intensity pixels of the input digital image.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (19)