# US 12,118,799 B2 · Method of selecting accident image using results of recognition of obstacle on road > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,118,799 B2 - Document ID: US12118799B2 - Status: Granted - Kind: B2 - Application number: 17/636,677 - Filed: 2021-12-29 - Published: 2023-07-13 - Granted: 2024-10-15 - Earliest claimed: 2020-12-31 - Assignees: Aimatics Co., Ltd. - Applicants: Aimatics Co., Ltd. - Inventors: Jin Hyuck Kim - Web page: https://patentgrep.com/patent/US12118799B2 ## CPC - G06V 20/58: Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/08: Learning methods - G06T 7/20: Analysis of motion (motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51) - G06T 7/50: Depth or shape recovery - G06T 7/73: using feature-based methods - G06V 10/25: Determination of region of interest [ROI] or a volume of interest [VOI] - G06V 10/764: using classification, e.g. of video objects - G06V 10/82: using neural networks - G07C 5/008: {communicating information to a remotely located station (transmission systems for measured values G08C)} - G07C 5/085: {using electronic data carriers} - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30261: Obstacle - G07C 5/0866: {the electronic data carrier being a digital video recorder in combination with video camera} - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - G07C: TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE ## Abstract The present disclosure relates to a method of selecting an accident image using the results of the recognition of an obstacle on a road, which can distinguish between an actual accident image and a fake accident image by previously recognizing an obstacle on a road while a vehicle travels and determining an impact event to have a low accident possibility, the impact event occurring in a section in which the vehicle goes over the obstacle, or suppressing the impact event, and can secure a space of a storage medium by deleting the fake accident image or prevent a data usage fee and unnecessary management expenses by blocking the transmission of the fake accident image to a remote cloud server. ## Claims ### Claim 1 A method of determining an accident image using recognition results of an obstacle on a road, by a blackbox or impact event generation apparatus installed in a vehicle, whether an image before or after impact detection timing is an actual accident image or a fake accident image, by using results of recognition of an obstacle on a road, the method comprising steps of: (a) obtaining, by a camera module of the vehicle, a photographed image of a road ahead of the vehicle, and detecting, by a processor provided in a blackbox or an impact event generation apparatus, an obstacle, which is located on the road ahead of the vehicle and whose impact is neglectable, from the photographed image; (b) determining, by the processor, a pass start time (T1) when the vehicle starts to pass through the obstacle and a pass end time (T2) when the vehicle completely goes over the obstacle; (c-1) detecting, by the processor, an occurrence of the impact event from an impact detection signal obtained by an impact detection sensor or a car area network (CAN) included the blackbox; (c-2) determining, by the processor, whether an impact event occurrence timing (Tc) is located between the pass start time (T1) and the pass end time (T2) when the impact event occurs; (c-3) determining, by the processor, the impact event to have a low accident possibility when the impact event occurrence timing (Tc) is located between the pass start time (T1) and the pass end time (T2); and (c-4) determining, by the processor, the impact event to have a high accident possibility when the impact event occurrence timing (Tc) is located before the pass start time (T1) or after the pass end time (T2). ### Claim 2 (depends on 1) The method of claim 1, wherein when the impact event has the low accident possibility, the processor does not store an image at the impact event occurrence timing (Tc) in a storage medium of the blackbox or the impact event generation apparatus or not transmit to a remote control server. ### Claim 3 (depends on 1) The method of claim 1, wherein when the impact event has the high accident possibility, the processor stores an image at the impact event occurrence timing (Tc) in an overwrite prohibition region of a storage medium of the blackbox or the impact event generation apparatus or immediately transmits to a remote control server. ### Claim 4 (depends on 1) The method of claim 1, wherein the obstacle is a speed bump or a pothole. ### Claim 5 (depends on 4) The method of claim 4, wherein the step (a) comprises steps of: operating, by a neural network learning unit comprising a plurality of convolution layers, at least one region of interest (RoI) through a region proposal network (RPN), detecting an object by pooling the RoI, calculating a class layer of the detected object, and then recognizing, as an obstacle, an object having a high coincidence ratio with a class layer of the speed bump or the pothole. ### Claim 6 (depends on 1) The method of claim 1, further comprising a step of detecting, by the processor, a distance (D) from the vehicle to the obstacle and a driving speed (v) of the vehicle after the step (a). ### Claim 7 (depends on 6) The method of claim 6, wherein the distance (D) to the obstacle is obtained using following equation: D = λ ⁢ h y - D c , wherein λ is a focal distance of the camera module, h is a height of the camera module from the ground, y is a vertical distance between an original point on a viewport photographed by the camera module and a bottom of a contour line of the obstacle, and Dc is a longitudinal distance from a front end of the vehicle to the camera module. ### Claim 8 (depends on 6) The method of claim 6, wherein the driving speed (v) is detected based on a value obtained by dividing a difference between the distances (D) to the obstacle at two different timings, at which the obstacle is recognized, by a time difference between the two timings. ### Claim 9 (depends on 6) The method of claim 6, wherein in the step (b), the pass start time (T1) and the pass end time (T2) are obtained using following equations, respectively: T ⁢ 1 = D y - Δ ⁢ T start T ⁢ 2 = ( D + d ) y + Δ ⁢ T end , wherein d is a length of the vehicle, ΔTstart is a margin for applying an offset to the obstacle pass start time, and ΔTend is a margin for applying an offset to the obstacle pass end time. ### Claim 10 (depends on 9) The method of claim 9, wherein the step (b) comprises steps of: (b-1) setting the pass start time (T1) and the pass end time (T2); (b-2) updating the pass start time (T1) by maintaining the impact event and detecting the distance (D) to the obstacle and the driving speed (v) again before the pass start time (T1) is reached; (b-3) suppressing the impact event after the pass start time (T1) is reached; (b-4) updating the pass end time (T2) by maintaining the suppression of the impact event and detecting the distance (D) to the obstacle and the driving speed (v) again before the pass end time (T2) is reached; and (b-5) restarting the impact event after the pass end time (T2) is reached.