Search

EP-4239588-B1 - METHOD AND APPARATUS FOR DETECTING BACKLIGHT OF IMAGE

EP4239588B1EP 4239588 B1EP4239588 B1EP 4239588B1EP-4239588-B1

Inventors

  • SONG, DONG HUI

Dates

Publication Date
20260506
Application Date
20220613

Claims (18)

  1. An apparatus (100) for detecting backlight of image, comprising: a memory (130) configured to store one or more instructions; and a processor (120) configured to execute the one or more instructions stored in the memory (130), wherein the processor (120): converts a pixel value for each of a plurality of image frames for a preset time period based on a first threshold, and blocks the pixel value into a plurality of block regions of a same size to generate a converted frame, generates brightness cumulative data based on the converted frame, determines, as a backlight candidate block region, a block region in which a block value of the block region is a maximum in the brightness cumulative data, classifies the plurality of block regions into first and second regions based on a position of the backlight candidate block region and calculates a backlight score of the image using a sum of block values for the block region of the first region and a sum of block values for the block region of the second region, and determines that the image is a backlight image based on the backlight score and a second threshold.
  2. The apparatus of claim 1, wherein the processor (120) determines severity of the backlight based on the sum of block values for the block region of the first region when the image is determined to be the backlight image.
  3. The apparatus of claim 1, wherein the processor (120) obtains a pixel value for a pixel of the frame and converts the pixel value to 0 when the pixel value is less than a preset threshold value.
  4. The apparatus of claim 3, wherein the pixel value is a pixel value related to brightness among pixel values related to a plurality of channels of the frame.
  5. The apparatus of claim 1, wherein the processor (120) divides the frame into a plurality of block regions having a preset size and calculates a block value of each block region, and the block value is an average value of pixel values for pixels corresponding to the block region.
  6. The apparatus of claim 1, wherein the bright cumulative data has a same number of block regions as a number of the block regions of the converted frame, and the block value of the block region of the brightness cumulative data is a value generated by summing block values for block regions at a same position corresponding to a position of the block region of the brightness cumulative data in the converted frame for each block region.
  7. The apparatus of claim 1, wherein the first region comprises the backlight candidate block region and a plurality of block regions at a position adjacent to the backlight candidate block region among a plurality of block regions included in the brightness cumulative data, and the second region includes remaining block regions included in the brightness cumulative data.
  8. The apparatus of claim 1, wherein the backlight score of the image is calculated based on a relative ratio between the sum of block values for the block region of the first region and the sum of block values for the block region of the second region.
  9. The apparatus of claim 2, wherein the processor (120) calculates intensity of the backlight based on the sum of block values for the block regions of the first region, and determines which level the image belongs to among two or more preset severity levels based on the intensity of the backlight.
  10. A computer-implemented method of detecting backlight of an image, comprising: converting a pixel value for each of a plurality of image frames for a preset time period based on a first threshold; blocking the pixel value into a plurality of block regions of a same size to generate a converted frame; generating brightness cumulative data based on the converted frame; determining, as a backlight candidate block region, a block region in which a block value of the block region is a maximum in the brightness cumulative data ; classifying the plurality of block regions into first and second regions based on a position of the backlight candidate block region and calculating a backlight score of the image using a sum of block values for the block region of the first region and a sum of block values for the block region of the second region, and determining that the image is a backlight image based on the backlight score and a second threshold.
  11. The method of claim 10, further comprising determining severity of the backlight based on the sum of block values for the block region of the first region when the image is determined to be the backlight image.
  12. The method of claim 10, wherein converting the pixel value for each of the plurality of image frames based on the first threshold comprises obtaining a pixel value for a pixel of the frame and converting the pixel value to 0 when the pixel value is less than a preset threshold value.
  13. The method of claim 12, wherein the pixel value is a pixel value related to brightness among pixel values related to a plurality of channels of the frame.
  14. The method of claim 10, wherein generating the converted frame comprises dividing the frame into a plurality of block regions having a preset size and calculating a block value of each block region, and the block value is an average value of pixel values for pixels corresponding to the block region.
  15. The method of claim 10, wherein the bright cumulative data has a same number of block regions as number of the block regions of the converted frame, and the block value of the block region of the brightness cumulative data is a value generated by summing block values for block regions at a same position corresponding to a position of the block region of the brightness cumulative data in the converted frame for each block region.
  16. The method of claim 10, wherein the first region comprises the backlight candidate block region and a plurality of block regions at a position adjacent to the backlight candidate block region among a plurality of block regions included in the brightness cumulative data, and the second region includes remaining block regions included in the brightness cumulative data.
  17. The method of claim 10, wherein the backlight score of the image is calculated based on a relative ratio between the sum of block values for the block region of the first region and the sum of block values for the block region of the second region.
  18. The method of claim 11, wherein determining the severity of the backlight includes calculating intensity of the backlight based on the sum of block values for the block regions of the first region and determining which level the image belongs to among two or more preset severity levels based on the intensity of the backlight.

Description

TECHNICAL FIELD The present disclosure relates to backlight detection, specifically, to a method and apparatus for detecting a backlight of an image generated by sunlight. BACKGROUND The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art. In autonomous driving and an advanced driver assistance system, ADAS, a vehicle uses various sensors, for example, a camera, radar, lidar, and ultrasonic sensors, to recognize the surrounding environment and a traveling situation. Among them, an image captured using a camera may provide image information on the surrounding environment to the vehicle, such as human visual information. With development of an autonomous driving technology, various image recognition systems for recognizing the surrounding environment using a camera are used in a vehicle. However, backlight may occur due to various light sources existing around the vehicle, and a backlight image taken under the backlight conditions is difficult to read properly because the brightness of some regions is distorted. In particular, due to backlight generated from strong sunlight, misrecognition or failure of recognition of the surrounding environment may occur in the image recognition system, which increases the risk of accidents in autonomous vehicles and decreases reliability of autonomous driving. In order to solve the problem of the backlight image, many methods for detecting and correcting a backlight image from a picture or a moving picture have been proposed. However, since conventional methods are designed to uniformly determine whether or not a backlight is present without considering the type and intensity of the light source, correcting the backlight image even in the case of a weak backlight capable of image recognition may cause unnecessary image distortion or an inaccurate detection of the backlight. US 2014/063288 A1 discloses an apparatus for detecting backlight of image, comprising: a memory configured to store one or more instructions; and a processor configured to execute the one or more instructions stored in the memory, wherein the processor creates a plurality of block regions of same size and generates brightness cumulative data; classifying the plurality of block regions into first and second regions and determining that the image is a backlight image based on the backlight score. KR 2017 0031982 A discloses vehicle monitoring system using image processing. SUMMARY A method and apparatus for detecting a backlight of image according to one embodiment generates brightness cumulative data based on a plurality of frames for a preset time period of an image and calculates a backlight score based on the brightness cumulative data to detect a backlight image by sunlight among various light sources. A method and apparatus for detecting a backlight of image according to one embodiment generates brightness cumulative data based on a plurality of frames for a preset time period of an image, and can determine severity of a backlight image based on the brightness cumulative data. The problems to be solved by the present disclosure are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description. According to one embodiment of the present disclosure, there is provided an apparatus for detecting backlight of image, comprising: a memory configured to store one or more instructions; and a processor configured to execute the one or more instructions stored in the memory, wherein the processor converts a pixel value for each of a plurality of frames of the image for a preset time period based on a first threshold value, and blocks the pixel value into a plurality of block regions of the same size to generate a converted frame, generates brightness cumulative data based on the converted frame, determines a block region in which a block value of the block region is the maximum in the brightness cumulative data as a backlight candidate block region, classifies the plurality of block regions into a first region and a second region based on a position of the backlight candidate block region and calculates a backlight score of the image using the sum of block values for the block region of the first region and the sum of block values for the block region of the second region, and determines the image as a backlight image when the backlight score is equal to or more than a preset second threshold. According to another embodiment of the present disclosure, there is provided a method for detecting backlight of an image, comprising: converting a pixel value for each of a plurality of frames of the image for a preset time period based on a first threshold value, and blocking the pixel value into a plurality of block regions of the same size to generate a converted frame; generating brightness cumulative data based on the converte