TWI270827B - Traffic lane recognition system and traffic lane deviation warning device with traffic lane recognition system and method of forming the same - Google Patents
Traffic lane recognition system and traffic lane deviation warning device with traffic lane recognition system and method of forming the same Download PDFInfo
- Publication number
- TWI270827B TWI270827B TW94121884A TW94121884A TWI270827B TW I270827 B TWI270827 B TW I270827B TW 94121884 A TW94121884 A TW 94121884A TW 94121884 A TW94121884 A TW 94121884A TW I270827 B TWI270827 B TW I270827B
- Authority
- TW
- Taiwan
- Prior art keywords
- vanishing point
- unit
- line
- image
- tracking
- Prior art date
Links
Landscapes
- Image Processing (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
Description
1270827 九、發明說明： . 【發明所屬之技術領域】 • 本發明是有關於一種車道辨識系統及具有該系統的車 道偏私警告裝置及其方法，特別是指一種以追蹤方式尋找 消失點及車道邊線的車道辨識系統及具有該系統的車道偏 移警告裝置及其方法。 【先前技術】 % 參閱圖1，中華民國專利證書號1228086揭露了 一種車 道偏移警不裝置，適用於一車輛，該裝置包含一車道視訊 貝汛擷取單το 401、一包括一標線偵測單元4〇3a與一車道 偏移判斷單元403b的車道偏移偵測單元4〇3及一警示單元 405 〇 該車道視訊資訊擷取單元401與該標線偵測單元4〇3a 電連接，並拍攝路面的原始影像，且將擷取的原始影像送 到該標線彳貞測單元403a。 _ 該標線偵測單元403a與該車道偏移判斷單元4〇3b電連 接，並將收到的原始影像進行影像處理，即利用一高通濾 • 波器，初步找出影像中線條明顯的部分，再加上一亮度下 • 限值，以得到更進一步的結果，且根據此結果計算搜尋區 域，並在搜尋區域内偵測車道標線，且將偵測結果送到該 車道偏移判斷單元403b。 該車道偏移判斷單元403b與該警示單元405電連接， 當該標線偵測單元403a在連續一段時間内偵測不到標線時 ，該車道偏移判斷單元403b判斷該車輛處於偏移狀態，並 1270827 送出一警示指令到該警示單元405。 該警示單元405在收到該警示指令時，會發出警示信 號來提醒駕駛人，以保障駕駛人的行車安全。1270827 IX. Description of the invention: [Technical field of the invention] The present invention relates to a lane recognition system and a lane bias warning device therefor, and a method thereof, in particular to finding a vanishing point and a lane edge in a tracking manner Lane recognition system and lane departure warning device having the same and method therefor. [Prior Art] % Referring to Figure 1, the Republic of China Patent No. 1228086 discloses a lane offset warning device suitable for a vehicle, the device comprising a lane video capture unit το 401, and a line detection The driving unit 4〇3a is connected to the lane offset detecting unit 4〇3 and the warning unit 405 of the one lane offset determining unit 403b, and the lane video information capturing unit 401 is electrically connected to the marking detecting unit 4〇3a. And taking the original image of the road surface, and sending the captured original image to the marking detection unit 403a. The reticle detecting unit 403a is electrically connected to the lane offset determining unit 4〇3b, and performs image processing on the received original image, that is, using a high-pass filter to initially find out the obvious line in the image. , plus a brightness limit value to obtain further results, and calculate a search area according to the result, and detect a lane marking in the search area, and send the detection result to the lane offset determining unit 403b. The lane offset determining unit 403b is electrically connected to the alert unit 405. When the reticle detecting unit 403a does not detect the reticle for a continuous period of time, the lane shift determining unit 403b determines that the vehicle is in an offset state. And 1270827 sends a warning command to the alert unit 405. When receiving the warning instruction, the warning unit 405 will issue a warning signal to remind the driver to ensure the driving safety of the driver.
由於該車道偏移警示裝置只對標線有較好的偵測效果 ’當車輛行駛在有標線的道路時’該裝置才能順利進行車 道偏移警示，當車輛行駛在無標線的道路時，該裝置會因 偵測效果太差幾乎失去效用，而且即使車輛行駛在有^線 的道路上’當旁邊有其它車輛行駛或標線附近有雜物時， 該裝置容易被干擾而找不到標線，產生誤判，因此不 用範圍受到限制，而且辨識度及準確度均偏低。 【發明内容】 奉發明之目的即在提供 么从 種旱道辨識系統，嗜 糸統以辨識及追蹤方式尋找消失點及車道邊線。 μ 本發明之另—目的即在提供—種具 車道偏移警告裝置，該裝置以辨識及追縱方式的 及車道邊線，並在車道偏移時發出警示。 *失點 本毛月之另目的即在提供一種車道辨識方 法可辨識及追蹤消失點及車道邊線。 ’邊方 種車道偏移警告方法， 邊線，並判斷車道是否 本發明之另一目的即在提供一 5亥方法可辨識及追縱消失點及車道 偏移。 巧平逼辨識系統的車 ，適用於輔助一車輛，並包八一、 偏移警告裝置 ^各一視訊擷取單亓、 識系統、一偏移判斷單元及一盤-抑 一車道辨 統 S不早元。該車道辨識系 1270827 辨識單元 包括-重點區域偵測單元、一影像處理單元 、一追蹤單元及一評估單元。 /視。fl擷取單凡拍攝路面的原始影像，並將擷取的原 始影像輸出。 吞亥重點區域彳自浪丨g ^ ^ 貝貝】早凡輸出一重點區域，並可適當調整 5亥重點區域的大小。 一忒影像處理單元與該視訊擷取單元及該重點區域偵測 早兀電連接，並從該重點區域情測單元接收該重點區域， 且將從該視訊擷取單开你 】並位於該重點區域中的原始影 像進饤邊緣摘測以得到_邊緣化影像。 該辨識單元與該影像處理單元電連接並在 處理單元收到的邊缝务旦冢 邊緣化衫像中辨識出至少-消失點後，將 所有通過該消失點的直線作像素梯度值累計 值最f的二直線即是該車道的左、右邊線。、 十 參 第一=跟早70與该影像處理單元及該辨識單元電連接， 一的影像先由該辨識單元辨識 、右邊線，然後該追蹤單^ μ 〇 天』及違左 ，由中心往外J 影像的消失點為中心 點，在從ρ像：定範圍内的像素作為可能的消失 ^在從“像處理單元收到的邊緣化影像中尋 月匕的消失點相對岸之^^ ^ 了 對應之可旎的左、右邊線，彡 的一組即是追縱出的消失點及左、右邊線。〇歲最強 單元與該辨識單元及該追縱單㈣ 辨識㈣識出該消失點及該左、右邊線，該評估單：:亥 收該辨識單元的辨識結果並輸出。當該 早疋接 κ早兀追蹤出該 1270827 消失點及該左、右邊線，該評 蹤結果，並評㈣追蹤單元 轉收該追料元的追 合理，該評估單元輸出目前收到的^確性。如果追縱結果 縱單元繼續追縱下一張影像的消失點=果二命令該追 ^估早％命令該追蹤單元停止追縱， 到的結果，且命令該辨識單 & -人收 左、右邊線。 辨識下一張影像的消失點及Since the lane offset warning device only has a good detection effect on the marking line, 'the vehicle can smoothly perform the lane offset warning when the vehicle is traveling on the road with the marking line, when the vehicle is traveling on the road without the marking line. The device will almost lose its effectiveness due to poor detection, and even if the vehicle is driving on a road with a line of 'when there are other vehicles driving or there are debris near the marking line, the device is easily disturbed and cannot be found. The marking line produces a false positive, so the range is not limited, and the degree of identification and accuracy are low. SUMMARY OF THE INVENTION The purpose of the invention is to provide a dryway identification system that uses the identification and tracking methods to find vanishing points and lane edges. Another object of the present invention is to provide a lane offset warning device that recognizes and tracks the lanes and lanes and alerts when the lanes are offset. * Lost point Another purpose of this month is to provide a lane recognition method to identify and track vanishing points and lane edges. The side lane warning warning method, the sideline, and the determination of whether the lane is another object of the present invention is to provide a 5 hai method identifiable and trace vanishing point and lane offset. A car that is easy to identify the system, suitable for assisting a vehicle, and includes a Bayi, offset warning device, a video capture unit, a recognition system, an offset determination unit, and a disc-one lane discrimination system. Not early. The lane recognition system 1270827 identification unit includes a focus area detection unit, an image processing unit, a tracking unit and an evaluation unit. / view. Fl takes the original image of the road surface and outputs the captured original image. The key area of Tenghai is from the wave g ^ ^ Beibei] early output of a key area, and the size of the 5 Hai key area can be adjusted appropriately. An image processing unit is electrically connected to the video capture unit and the focus area, and receives the key area from the focus area sensing unit, and the video is taken from the video and is located at the focus The original image in the area is edged to obtain an edged image. The identification unit is electrically connected to the image processing unit, and after recognizing at least the vanishing point in the edge stitching shirt image received by the processing unit, all the straight lines passing through the vanishing point are used as the cumulative value of the pixel gradient value. The two straight lines of f are the left and right lines of the lane. The tenth reference first=the early 70 is electrically connected to the image processing unit and the identification unit, and the image of one is first recognized by the identification unit, the right line, and then the tracking unit is μ天〇 and the left is left, from the center to the outside The vanishing point of the J image is the center point, and the pixel from the ρ image: the range is possible to disappear. ^ From the marginal image received by the processing unit, the vanishing point of the moon is opposite to the shore ^^ ^ Corresponding to the left and right lines, the group of 彡 is the vanishing point and the left and right lines. The oldest unit and the identification unit and the tracking list (4) identify (4) the vanishing point and The left and right lines, the evaluation sheet:: the recognition result of the identification unit is outputted and outputted. When the early 疋 κ 兀 traces the 1270827 vanishing point and the left and right lines, the evaluation result is evaluated. (4) The tracking unit transfers the tracking element's chasing reason, and the evaluation unit outputs the current confirmation. If the tracking result unit continues to track the vanishing point of the next image = fruit 2 command the tracking Commanding the tracking unit to stop tracking, to the knot And the identification of a single command & - person to the left and right by-line identification of the vanishing point and the next image.
〇偏移判斷單元與該評估單 單元輸出的/ 接，並接收該評估 天‘占及左右邊線，當判斷出車道偏蒋日#， 即連續複數張影像的左、右邊 # 夺 方向燈，送出-警示信號。不疋洛在車頭兩旁且未打 單元與該偏㈣斷單元電連接，並在收到該警 4錢’可依該警示信號通知該車輛的駕敬人。 而本發明車道偏移警告方法包含以下步驟： (A) 擷取路面的原始影像； (B) 將位於—重點區域中的原始影像進行邊緣偵測以 得到一邊緣化影像； (C) 從該邊緣化影像中辨識出至少—消失點，並將所 有通過該消失點的直線作像素梯度值累計，其中，累計值 最大的一直線即是該車道的左、右邊線； (D) 根據辨識出的消失點及左、右邊線判斷是否有車 道偏移’即連續複數張影像的左、右邊線不是落在車頭兩 旁且未打方向燈，如果是，跳到步驟（E)，否則，跳到步 驟（F ); !27〇827 (E )輸出一警示信號； 、 ⑺擷取下-張路面的原始影像； 、 〜（G)將㈣該重點區域中的原始影像進行邊緣制以 传到該邊緣化影像； (H) 以上—張影像的消失點為中心、，由中^往外，依 序Μ特定範圍内的像素作為可能的消失點，在該邊緣化影 像中尋找與該可能的消失點相對應之可能的左、右邊線， • 丨中’信號最強的-組即是追蹤出的消失點及左、右邊線 (I) 評估追蹤的正確性，如果追蹤結果合理，跳到步. 驟（J )，否則，跳到步驟（Κ ); ⑴根據目前追蹤出的消失點及左、右邊線判斷是否The 〇 offset judging unit and the output of the evaluation unit unit are connected, and receive the evaluation day 'occupies the left and right side lines, and when it is judged that the lane is biased toward the Jiang #, that is, the left and right sides of the continuous plurality of images are steered by the direction lights, and are sent out. - Warning signal. If the door is not on the front of the car and the unit is not connected to the partial (four) disconnected unit, and the policeman receives the money, the driver can be notified according to the warning signal. The lane departure warning method of the present invention comprises the following steps: (A) capturing the original image of the road surface; (B) performing edge detection on the original image located in the key area to obtain a marginal image; (C) from the At least the vanishing point is recognized in the edged image, and all the straight lines passing through the vanishing point are accumulated as pixel gradient values, wherein the straight line with the largest cumulative value is the left and right lines of the lane; (D) according to the identified The vanishing point and the left and right lines determine whether there is a lane offset. That is, the left and right lines of the continuous plurality of images do not fall on both sides of the front and do not hit the direction lights. If yes, skip to step (E), otherwise, skip to the step. (F); !27〇827 (E) outputs a warning signal; (7) extracts the original image of the next road surface; , (G) (4) edges the original image in the key area to pass to the edge (H) Above the vanishing point of the image is centered, and from the middle to the outside, the pixels in the specific range are sequentially used as possible vanishing points, and the edged image is searched for the possible vanishing point. Corresponding possibility Left and right lines, • The 'strongest signal' in the middle is the tracking vanishing point and the left and right lines (I) to assess the correctness of the tracking. If the tracking result is reasonable, skip to step (J), otherwise , skip to the step (Κ); (1) judge whether it is based on the current vanishing point and the left and right lines
有車道偏移’如果是，跳到步驟（Ε)，否則，跳到步驟（F ); (K)根據上-次辨識或追縱出的消失點及左、右邊線 • 判斷是否有車道偏移，如果是，跳到步驟（L)，否則，跳 到步驟（Μ ); ‘ (L )輸出該警示信號；及 ‘ （Μ)調整該重點區域的大小，並跳到步驟（Α)。 【實施方式】 有關本發明之前述及其他技術内容、特點與功效，在 以下配合參考圖式之一個較佳實施例的詳細說明中，將可 清楚地呈現。 參閱圖2,轉明具有車道辨識系統的車道偏移警告裝 9 1270827 置適用於輔助一車輛’並包含一視訊掏取單元丄、一車道辨 識系統2、-偏移判斷單元3及—㈣單U。該車道辨識 糸統包括-影像處理單A 21、—重點區域偵測單元Μ、一 辨識單元23、-追蹤單元24及—評估單元25。 該視訊擷取單元1與該影像處理單元21電連接，並拍 攝路面的原始影像，1脸私 理單元21。 將掏取的原始影像傳送到該影像處 、該重點區域_單元22輪出-重點區域，而該重點區 或身又疋才曰車頭别方的部份影像，當該重點區域中的影像 被阻擋，導致辨識資訊不㈣，該重點區域偵測單元22可 適當加大該重點區域的大小。 心像處理早凡21與該重點區域㈣單元22、該辨識 早兀23及該追縱單元24電連接，並從該重點區域痛測單 元22接收該重點區域， 且將從泫視訊擷取單元丨收到並位 於該重點區域中的原始影像進行邊緣制以得到 並將該邊緣化影像送到該辨識單元23及該追縱單元 」原始’V像加以處理可則貞财標線及無標線的道路 間化影像資訊，加快後續的演算速度。 ^知典型的邊緣制方法n制… 里維特（Prewitt )、、宫曾、、也 。 /、异法’利用影像頻率的不連續性，對 法，：::處::元21的邊緣偵測可以是使用索貝爾演算 刀別將-水平方向空間渡波器、一對角線方向空間 10 1270827 濾波器ίί2及一反對角線方向空問 P(x，y)作迴旋積（C〇nv〇luti〇n)心、；裔％與該原始影像 的大J ^ r, , , 迷計算像素梯度值N(x，y) 的大小（Amphtude) |N(x，y)卜以 角線方向的物件邊緣。計算方以财平、對角線及反對 "-1 0 Γ H' = - 2 0 2 一 1 0 1_ '2 1 h2 = 1 0 -1 0 -1 - 2 '0 1 2 H3 = -1 0 1 一 2 一 1 0 % (x，少)'结邮，料少力) 灿，鳥ΉThere is lane offset 'if yes, skip to step (Ε), otherwise, skip to step (F); (K) according to the last-time identification or chasing vanishing point and left and right lines • Determine whether there is lane deviation Move, if yes, skip to step (L), otherwise, skip to step (Μ); '(L) output the warning signal; and '(Μ) adjust the size of the focus area and jump to step (Α). The above and other technical contents, features, and advantages of the present invention will be apparent from the following detailed description of the preferred embodiments. Referring to Figure 2, the lane offset warning device 9 1270827 with the lane recognition system is adapted to assist a vehicle' and includes a video capture unit, a lane recognition system 2, an offset determination unit 3, and a (four) list. U. The lane recognition system includes an image processing unit A 21, a key area detecting unit Μ, an identifying unit 23, a tracking unit 24, and an evaluation unit 25. The video capturing unit 1 is electrically connected to the image processing unit 21, and takes an original image of the road surface, a face-space unit 21. The captured original image is transmitted to the image, the key area _ unit 22 is rotated-key area, and the key area or the body is only part of the image of the front of the vehicle, when the image in the focus area is Blocking, causing the identification information not (4), the focus area detecting unit 22 can appropriately increase the size of the key area. The heart image processing system 21 is electrically connected to the key area (4) unit 22, the identification early 23 and the tracking unit 24, and receives the key area from the key area pain detecting unit 22, and the video capturing unit is The original image received and located in the focus area is edge-made and sent to the identification unit 23 and the tracking unit "original" V image for processing, and the financial line and the unmarked line are processed. Inter-road image information to speed up the subsequent calculations. ^ Know the typical method of marginal system n... Prewitt, Palace, and also. /, the different method 'use the discontinuity of the image frequency, the law, ::::: 21 edge detection can be using Sobel calculus knife - horizontal direction space ferrite, a diagonal direction space 10 1270827 Filter ίί2 and an anti-angular direction space P(x, y) for the gyro product (C〇nv〇luti〇n) heart, the %% and the original image of the large J ^ r, , , The size of the pixel gradient value N(x,y) (Amphtude) |N(x,y) is the edge of the object in the angular direction. Calculate the currency, diagonal and opposition "-1 0 Γ H' = - 2 0 2 -1 0 1_ '2 1 h2 = 1 0 -1 0 -1 - 2 '0 1 2 H3 = -1 0 1 - 2 - 1 0 % (x, less) 'send mail, less energy" Can, bird
綠矽琢爽埋皁兀21的邊续居、日丨L m 〕邊緣偵測也可以是使用普里維特 决异法，为別將一水平方&处The edge of the green scented saponin 21, the 丨L m 〕 edge detection can also be used to use the Privet ruin method, for a horizontal party &
>卞万向工間濾波器H 空間濾波器H2及一反對备绐^ 丁用踝方向 ㈣角線方向空間濾波 像P(x，y)作迴旋積，並計算像 /原始〜 冢素梯度值N(x，y)的大小|N(x，y) 11 1270827 ’以強調水平、對角線及反對角線方向的物件邊緣。計算 方式如下：> 卞 universal work filter H spatial filter H2 and an anti-preparation 丁 丁 踝 ( direction (four) angular direction spatial filtering image P (x, y) for the gyro product, and calculate the image / original ~ 冢 gradient The value of the value N(x,y)|N(x,y) 11 1270827 'to emphasize the edge of the object in the horizontal, diagonal, and diagonal directions. The calculation is as follows:
Νι{χ^)=Σ^ Σ^ι (^j)p(x - - j) Ϊ=-1j=-\ N2{^y)=^HH2 J)pix -hy-j) /=-1 y=—1 # NAx，y)=HHA f)P[x - i，y - j) /=-1y=-l • \N(^ y] = +1^2 y)+1^3 3^®/3 - 由於典型的邊緣偵測方法容易受到雜訊的干擾，同時 也會放大高頻訊號，為避免上述缺點，該影像處理單元21 的邊緣债測也可以是使用習知型態學（Morphological )演 算法，利用原始影像P(x，y)及一遮罩A所含蓋之像素的最 小值min(A)計算像素梯度值大小 Τν(χ，3；) = Ρ(χ，3；)_ηιίη«〇 12 1270827 口多閱圖3，该辨識單元23利用素描常用的透視圖法， 只要找出二度平面中的消失點Q位置，即可還原三度空間 中的平仃線。因為在三度空間中相互平行的所有物件邊緣 、土如又有與視覺水平線M呈垂直或平行，其延伸線必在 ，方^於J5J點’該點即是消失點Q。而車道邊緣的切線就 疋平仃線’即使是彎彎曲曲的車道也是一樣，其邊緣的切 線必通過該消失點Q。Νι{χ^)=Σ^ Σ^ι (^j)p(x - - j) Ϊ=-1j=-\ N2{^y)=^HH2 J)pix -hy-j) /=-1 y =—1 # NAx,y)=HHA f)P[x - i,y - j) /=-1y=-l • \N(^ y] = +1^2 y)+1^3 3^® /3 - Since the typical edge detection method is susceptible to noise interference and also amplifies the high frequency signal, in order to avoid the above disadvantages, the edge processing of the image processing unit 21 may also use the conventional morphology (Morphological). The algorithm calculates the pixel gradient value Τν(χ,3;) = Ρ(χ,3;) using the original image P(x,y) and the minimum value min(A) of the pixel covered by a mask A. _ηιίη«〇12 1270827 More than FIG. 3, the identification unit 23 uses the perspective method commonly used in sketching to restore the flat line in the three-dimensional space by finding the position of the vanishing point Q in the second plane. Because all the edges of the objects parallel to each other in the three-dimensional space are perpendicular or parallel to the visual horizontal line M, the extension line must be at the point of J5J point, which is the vanishing point Q. The tangent to the edge of the lane is the same as the curved lane. The tangent to the edge must pass through the vanishing point Q.
>閱圖2 ’違辨識單元23與該重點區域谓測單元22、 該追蹤單元24及古岁士承/士留-1 r % 士 〇〇 μ ϋ平估早兀25電連接，並在從該影像處 理單元21收到的邊緣化影像中辨識出至少一消失點後，將 所有通過該消失點的直線作像素梯度值累彳，其中，累計 值最大的—直線即疋該車道的左、右邊線。該辨識單元Μ 將辨識出的消失點傳送到該重點區域偵測單元22及該追蹤 皁元24，將辨識出的消失點及左、右邊線傳送到該評估單 在本實施例中，該辨識單元23是利用霍氏轉換（ Hough Transform)及霍氏反轉換從該邊緣化影像中辨識出 =失點。霍氏轉換是將直角座標平面上的直線轉換成極 座標平面上的點，以法線的距離及法線與X軸之間的夾角 表示。 門 該辨識單元23先將收到的邊緣化影像分成複數小區域 ，再對-小區域中的每一像素作霍氏轉換，以在極 面上產生複數曲線’當該等曲線有交點時，即在直角二， 平面上有直線，找出該等曲線的交點，再換下-小區域: 13 1270827 續作霍氏轉換’直到所有小區域都做完轉換。 此時，該辨識單元23將 ，以在直角朗㈣反轉換 通μ❹或〜 生稷數直線’其十，有最多直線 通過的點即為該消失點。因為影像處 的消失點可能不只一個。今辨，， 的决差’找到 β辨識早几23將通過找到的該消 :最==點的直線作像素梯度值累計，其中，累計 2取大的二直線即是辨識出的左、右邊線，而該 交點即是辨識出的消失點。 運琢曰7 該辨識單元23利用二产★門由从7 y 度工間中的任何物件辨識該消失 二1::而且將收到的邊緣化影像分成複數小區 高車道的辨識度。避免其匕車辆或雜物的干擾，提 ::重：區域偵測單元22接收該辨識單元23辨識出的 =點’㈣_單元23有找料失點幻肖失點 夠尚，該重點區域轉元22在該辨識翠元23下次作動 月==點區域縮小為原始影像大小的1/3，當該辨識單元 J單點或消失點的集中度太低，該重點_ 象:識單元23下次作動前將該重點區域擴大 為原始⑹像大小的2/3 ’其餘時候該重點區域維持不變。 该追縱單元24與該評估單元25電連接。第— 的影像先由該辨識單元23 又刖 “ 平 辨為出_失點及該左、右邊線 ’以後该追縱單元24以上—張影像的消失點為中心，由中 序以特定範圍内的像素作為可能的消失點，在 技麟里單元21收到的邊緣化影像中尋找與該可能的 14 1270827 消失點相對應之可能的左、右邊線，其中，信號最強的一 :即是追㈣㈣失點及左、右邊線。該追縱單元24將追 舨出的消失點及左、右邊線傳送到該評估單元乃。 ^本實施例中，該追縱單元24尋找可能之左、右邊線 =方式如下：從可能㈣失點往車頭中心畫—直線，將該 線在〇度方向旋轉’並對該直線作像素梯度值累計，其 中，累計值最大的直線為可能的左邊線，將該直線往18〇 度方向旋轉，並對該直線作像素梯度值累計，其中，累計 值最大的直線為可能的右邊線。 該追縱單元24判斷追蹤出之消失點及左、右邊線的方 式士下.利用第1組可能之左邊線的像素梯度值累計队及 可能之右邊線的像素梯度值累計叫，計算可能之消失點& 的信號XCQ^I^Ir，其中，信號最大的一組即是追縱出 的消失點及左、右邊線。 仰，亥汗估單疋25肖該偏移判斷單a 3電連接，當該辨識 單元23辨識出該消失點及該左、右邊線，該評估單元二 接收該辨識單元23的辨識結果並傳送到該偏移判斷單元3 :該辨識結果包括該左、右邊㈣度1該追蹤單元Μ追 知出该消失點及該左、右邊線’該評估單元25接收該追縱 單元24的追蹤結果，肖追蹤結果包括該左、右邊線角度。 該追蹤單元並根據下式計算出—角度差異及一位置差 異，以評估該追蹤單元24追蹤的正確性·· Δ W (卜1) ♦ (η) 一〜(7卜q 15 1270827 其中，δθ是該角度差異，aq是該位置差異，θ^(η)、 心⑻及Q⑻分別是目前收到的左、右邊線角度及消失點位 置，而eL(n-l)、0R(n-l)及Q(n·!)分別是上一次收到的左、 右邊線角度及消失點位置。 汝果δ亥角度差異及該位置差異均小於特定值，表示追 蹤結果合理，該評估單元25輸出目前收到的追蹤結果到該 偏移判斷單元3，並命令該追蹤單元24繼續追蹤下一張影 像的消失點及左、右邊線。 如果該角度差異大於特定值且該位置差異小於特定值 :表示追蹤結果不合理，該評估單元25命令該追蹤單元Μ 钐止追sk，並輸出上一次收到的結果到該偏移判斷單元3， 且命令該辨識單元23辨識下一張影像的消失點及左、右 線。 =如果該位置差異大於特定值，表示追蹤出的消失點非 最佳’該評估單元25命令該追縱單元24以目前追縱出的 肩失點為巾心，繼續對目前影像追縱該消失點及該左、右 j線，當該追蹤單元24對目前影像重覆追蹤超過一定次數 k ’ Μ估單it 25命令該追蹤單元24停止追縱，並輸出 一』收到的追縱結果到該偏移判斷單& 3，1命令該辨識單 tl 23辨識下―張影像的消失點及左、右邊線。 、j偏移判斷單元3與崎示單元5電連接，並接收該 平估單元25輸出的消失點及左、右邊線，當判斷出車道偏 移或：車道偏移時，送出一警示信號到該警示單元5。車道 偏移疋.曰連績複數張影像的左、右邊線不是落在車頭兩旁 16 1270827 且未打方向燈。準車道偏移是指左、右邊線落在車頭兩旁 =疋方向盤的角度感應器測得的行車方向與車頭中心指 向消失點的道路方向相差一特定值。 4 '不早^ 5在收到該警示信號後，可依該警示信號 通知該車輛的駕駛人。 參閲圖4 ’本發明車道偏移警告方法包含以下步驟. 且將η該視訊擷取單元1拍攝路面的原始影像， ;。取的原始影像傳送到該影像處理單元2卜 步驟議是該影像處理單元21將位於該重點區域中的 …衫像進行邊緣_以得到該邊緣化影像。 邊緣二Π*是該辨識單元23接收該邊緣化影像，並將該 逯緣化影像分成複數小區域。 該辨識單元23對—小_的每—像素作 =，以在極座標平面上產生複數曲線，當該等曲線 交二在㈣座標平面上有直線，找出該等曲線的 都做完轉再:。下一小區域繼續作霍氏轉換，直到所有小區域 轉換步!：Π識單元23將所有找到的交點作霍氏反 、 直角座標平面上產生複數直線，其中，有,夕 ^線通過的點即為該消失點。因為影像處 了 找到的消失點可能不只—個。 的决差 等、、肖=Γ辨識單元23將通過找到的該消失點或該 像素梯度值累計’其中，累計值最大的 疋‘出的左、右邊線，而該二邊線的交點即是 17 1270827 點。該評估單元25接收該辨識單元。的辨 並傳送到該偏移判斷單元3。 斷Η * Μ 偏移判斷單元3根據收到的辨識結果判 :否有車道偏移或準車道偏移。如果是，㈣步驟綱， 否則’跳到步騍8〇9。 步驟__是該偏移判斷單元3送出該警示信號到該警 人。疋5’且該警示單元5依㈣示信號壯該車柄的駕驶 且將押Γ _7^視訊#|取單元1拍攝路㈣原始影像， ㈣的原始影像傳送到該影像處理單元21。 〜 G疋'^像處理單元21將位於該重點區域中的 ''象進行邊緣_以得到該邊緣化影像。 心，2川是該追縱單元24以前一張影像的消失點為中 失 。在外’依序以特定範圍内的像素作為可能的消 ^ ’在從該影像處理單元21收到的邊緣化影 .% ^ W之了肊的左、右邊線，其中，信號 =的-組即是追蹤出的消失點及左、右邊線。該追縱單 凡24將追蹤結果傳送到該評估單元〜 =812是該評估單元25根據目前影像的左、右邊線 點I::失:位置及上一張影像的左、右邊線角度及消失 ’、、 ，計算該角度差異及該位置差異。 步驟M3是該評估單以5評估該位置差異是否合理， 跳到步驟㈣’否則，跳到步 驟 814 〇 18 127〇827 步驟814是該評仕w 一 目前影像重覆追縱超過:二：判蹤单元24是否對 否則，跳到步驟816。數果疋’跳到步驟815, 步驟8'是該評估單以5命令該追蹤單元 二並輸出目則收到的追蹤結果到該偏移判斷單元/追 識單元23辨識下-張影像的消失點及左右二命 跳到步驟820。 右邊線。 步驟816是該追跑s 1 心，由中心往外，tr 追蹤出的消失點為中> Read Figure 2 'Illegal identification unit 23 and the key area predicate unit 22, the tracking unit 24 and the ancient ages/Shishang-1 r % 士〇〇μ ϋ 估 兀 兀 25 25 25 25 25 25 25 25 25 After recognizing at least one vanishing point from the edged image received by the image processing unit 21, all the straight lines passing through the vanishing point are accumulated as pixel gradient values, wherein the line with the largest cumulative value is the left of the lane. , the right line. The identification unit 传送 transmits the identified vanishing point to the key area detecting unit 22 and the tracking soap element 24, and transmits the identified vanishing point and the left and right lines to the evaluation sheet. In the embodiment, the identification Unit 23 identifies the missing point from the edged image using Hough Transform and Hawker inverse transform. The Hall's transformation is the conversion of a straight line on a Cartesian coordinate plane to a point on the polar coordinate plane, expressed as the distance between the normal and the angle between the normal and the X-axis. The door recognition unit 23 first divides the received edged image into a plurality of small regions, and then performs a Holstein transformation on each pixel in the small region to generate a complex curve on the polar surface. When the curves have intersections, That is, at right angle two, there is a straight line on the plane, find the intersection of the curves, and then replace the small area: 13 1270827 Continue to perform the Hollock conversion until all the small areas have been converted. At this time, the identification unit 23 sets the point at which the straight line is reversed (four) or the number of the line is 'n', and the point at which the most straight line passes is the vanishing point. Because there may be more than one vanishing point at the image. In this case, the difference of ', find the β identification early 23 will be through the found: the == point of the line as the pixel gradient value accumulation, where the cumulative 2 takes the big two straight line is the identified left and right Line, and the intersection is the identified vanishing point. The identification unit 23 recognizes the disappearance by any object in the 7 y degree using the second production door: and divides the received marginalized image into the recognition of the plurality of cells in the high lane. To avoid the interference of the vehicle or the debris, the following:: The area detecting unit 22 receives the = point identified by the identification unit 23 (4) _ unit 23 has the missing point of the missing point of the material. The area transfer element 22 is reduced to 1/3 of the original image size in the next activation month== dot area of the identification uiyuan 23, and the concentration of the single point or vanishing point of the identification unit J is too low, the focus _ The unit 23 expands the focus area to 2/3 of the original (6) image size before the next operation. The remaining key area remains unchanged. The tracking unit 24 is electrically connected to the evaluation unit 25. The image of the first is first determined by the identification unit 23 as "the point of the _ loss point and the left and right lines" after the tracking unit 24 is above the vanishing point of the image, and the middle order is within a specific range. As a possible vanishing point, the pixel is found in the marginalized image received by the unit 21, and the possible left and right lines corresponding to the possible 14 1270827 vanishing point, wherein the signal is the strongest: (4) (4) The missing point and the left and right lines. The tracking unit 24 transmits the vanishing point and the left and right lines of the tracking to the evaluation unit. In this embodiment, the tracking unit 24 searches for the possible left and right sides. Line = the way is as follows: from the possible (four) loss point to the center of the front of the car - straight line, the line is rotated in the direction of the twist 'and the line is the pixel gradient value accumulation, wherein the line with the largest cumulative value is the possible left line, will The straight line rotates in the direction of 18 degrees, and the line is integrated with the pixel gradient value, wherein the line with the largest integrated value is the possible right line. The tracking unit 24 determines the way to track the vanishing point and the left and right lines. Shixia. Using the pixel gradient value of the left-hand line of the first group, the pixel gradient value of the cumulative team and possibly the right-hand line is accumulated, and the signal XCQ^I^Ir of the possible vanishing point & is calculated, wherein the largest group of signals is The vanishing point and the left and right lines of the chase are traced. The elevation is determined by the offset, and the offset unit judges the single a 3 electrical connection. When the identification unit 23 recognizes the vanishing point and the left and right lines, the evaluation is performed. The unit 2 receives the identification result of the identification unit 23 and transmits the identification result to the offset determination unit 3: the identification result includes the left and right (four) degrees 1 the tracking unit Μ catches up the vanishing point and the left and right lines 'the evaluation The unit 25 receives the tracking result of the tracking unit 24. The tracking result includes the left and right line angles. The tracking unit calculates an angle difference and a position difference according to the following formula to evaluate the correctness of the tracking unit 24 tracking. ·· Δ W (卜1) ♦ (η) 一～(7卜q 15 1270827 where δθ is the difference in angle, aq is the difference in position, θ^(η), heart (8) and Q(8) are currently received Left and right line angles and vanishing point positions, and eL(nl) 0R(nl) and Q(n·!) are the left and right line angles and vanishing point positions received last time respectively. The difference between the angles of δ海 angle and the position difference are smaller than the specific value, indicating that the tracking result is reasonable. The evaluation unit 25 outputs the currently received tracking result to the offset determination unit 3, and commands the tracking unit 24 to continue tracking the vanishing point and the left and right lines of the next image. If the angle difference is greater than a specific value and the position difference Less than a specific value: indicating that the tracking result is unreasonable, the evaluation unit 25 instructs the tracking unit to track sk, and outputs the last received result to the offset determining unit 3, and instructs the identifying unit 23 to recognize the next one. The disappearance point of the image and the left and right lines. If the difference in position is greater than a specific value, it indicates that the disappearance point of the tracking is not optimal. The evaluation unit 25 instructs the tracking unit 24 to use the shoulder loss point currently being traced. Heart, continue to track the vanishing point and the left and right j lines of the current image, when the tracking unit 24 repeatedly tracks the current image for more than a certain number of times, the tracking unit 24 commands the tracking unit 24 to stop tracking, and lose A "追縱 results received to determine the offset single & 3,1 command to identify the single identification tl 23 - vanishing point image but left and right line. The j offset determination unit 3 is electrically connected to the display unit 5, and receives the vanishing point and the left and right lines output by the evaluation unit 25, and when determining the lane offset or the lane offset, sends a warning signal to The warning unit 5. The lanes are offset. The left and right lines of the multiple images are not on the front of the car. 16 1270827 and no direction lights. The quasi-lane offset means that the left and right lines fall on both sides of the front of the vehicle. = The direction of the steering sensor measured by the angle sensor is different from the direction of the road leading to the vanishing point. 4 'Not early ^ 5 After receiving the warning signal, the driver of the vehicle can be notified according to the warning signal. Referring to FIG. 4, the lane departure warning method of the present invention includes the following steps: and the video capturing unit 1 captures the original image of the road surface. The original image is transferred to the image processing unit 2, and the image processing unit 21 performs an edge image of the shirt image located in the focus area to obtain the edged image. The edge binogram* is that the recognition unit 23 receives the edged image and divides the edged image into a plurality of small regions. The identification unit 23 performs a = for each pixel of the small_ to generate a complex curve on the polar coordinate plane. When the curve intersects with a straight line on the (four) coordinate plane, it is found that the curves are all completed: . The next small area continues to perform the Hollock conversion until all the small area conversion steps!: The identification unit 23 generates all the intersections on the plane of the Holstein's inverse and right angle coordinates, where there is a point through which the line passes This is the vanishing point. Because the disappearance point found in the image may not only be one. The difference, etc., the Γ=Γ recognition unit 23 will accumulate the left and right lines of the 累计' where the accumulated value is the largest by the found vanishing point or the pixel gradient value, and the intersection of the two lines is 17 1270827 points. The evaluation unit 25 receives the identification unit. The discrimination is transmitted to the offset judging unit 3. Breaking Η * 偏移 The offset determining unit 3 judges based on the received recognition result that there is a lane offset or a lane change. If yes, (4) Steps, otherwise, go to Step 8〇9. Step __ is that the offset determination unit 3 sends the warning signal to the policeman.疋5' and the warning unit 5 drives the handle of the handle according to the signal of (4) and will transfer the original image of the unit (4) to the original image, and the original image of (4) is transmitted to the image processing unit 21. The image processing unit 21 performs an edge image of the image located in the focus area to obtain the edged image. Heart, 2 Sichuan is the disappearance point of the previous image of the memorial unit 24 is medium loss. In the outer 'sequentially, pixels in a specific range are taken as possible. In the left and right lines of the edged shadow. % ^ W received from the image processing unit 21, wherein the signal = the group is It is the vanishing point and the left and right lines that are tracked out. The tracking unit 24 transmits the tracking result to the evaluation unit~=812, the evaluation unit 25 according to the left and right line points of the current image I::mis: position and the left and right line angles of the previous image and disappear ',,, calculate the difference in angle and the difference in position. Step M3 is to evaluate whether the position difference is reasonable by 5, skip to step (4) 'Otherwise, skip to step 814 〇18 127〇827 Step 814 is the reviewer w. The current image is repeated more than: 2: Judgment If the trace unit 24 is otherwise, skip to step 816. The number 疋 'jumps to step 815, step 8' is that the evaluation order 5 commands the tracking unit 2 and outputs the tracking result received by the target to the offset determining unit/following unit 23 to recognize the disappearance of the next image Point and left and right jump to step 820. Right line. Step 816 is to chase the s 1 heart, from the center to the outside, the disappearance point traced by tr is medium
卜依序以特定範圍内的像素作A ;點’在從該影像處理單元…的邊緣化影像;= =可能的消失點相對應之可能的左、右邊線中= 敢強的一組即是追蹤出的消失點及左、右邊線，。\、自就 70 24將追蹤結果傳送到該評估單元25。 "蹤早 角产Π Γ7广該評估單元2 5根據目前收到的左、右邊線 點位置及上-次收到的左、右邊線角度及消失 置，计舁該角度差異及該位置差異。跳到步驟813。 是該評估單元25評估該纽差異是否合理， =何㈣。叫，跳_.心跳到步 "步驟819是該評估單元25命令該追縱單元24停止追 跟’並輸出上-次收到的結果到該偏移判斷單元3，且命令 该辨識單元23辨識下—張影像的消失點及左、右邊線。7 步驟820是該偏移判斷單元3根據收到的結果判斷是 否有車道偏移或準車道偏移^果是，關步驟821，否則 19 !27〇827 ’跳到步驟822。 . 步.驟821是該偏移判斷單元3送出該警示信號到料 、 示單元5 ’且該警示單元5依該警示信號通知該車辅的駕駛 人0 _ V驟822疋该重點區域偵測單元22接收該辨識單元23 辨識出的消失點，當有找到消失點且消失點的集中度夠高 ，將該重點區域縮小為原始影像大小的1/3，當沒有找到消 • 《點或消失點的集中度太低，將該重點區域擴大為原始影 像大J的2/3，其餘時候該重點區域維持不冑，並將該重點 區域傳送到該影像處理單元21。跳到步驟8〇1。 ▲步驟823是該評估單元25輸出目前收到的追蹤結果到 該偏移判斷單元3，並命令該追蹤單元24繼續追縱下一張 影像的消失點及左、右邊線。 步驟824該偏移判斷單& 3根據收到的追蹤結果判斷 是否有車道偏移鱗車道偏移。如果是，關㈣綱，否 | 則，跳到步驟809。 綜上所述，本發明車道辨識系統及具有該系統的車道 . ㈣警告裝置及其方法先將原始影像作邊緣制，並以透 ㈣法辨識出邊緣化影像中的消失點及左、右邊線，加上 重點區域可以適時加大，因此無論車輛行駛在有標線或無 標線的道路上，甚至有其它車輛或雜物干擾，都可以辨識 出車道，大大地提高車道的辨識度，且高車道辨識度可以 避免車道偏移的决判，也大大地提高車道偏移判斷的準確 度。而且當辨識出邊緣化影像中的消失點及左、右邊線後 20The order of the pixels in the specific range is A; the point 'in the marginalized image from the image processing unit...; = = the possible vanishing point corresponding to the left and right lines = the strong group is Track out the vanishing point and the left and right lines. The self-contained 70 24 transmits the tracking result to the evaluation unit 25. " trace early angle Π 广 7 wide The evaluation unit 2 5 according to the current left and right line position and the left and right line angles and disappearance received, calculate the angle difference and the difference . Skip to step 813. It is the evaluation unit 25 that evaluates whether the difference is reasonable, = (4). Step 819 is that the evaluation unit 25 instructs the tracking unit 24 to stop following and output the result of the last-time reception to the offset determination unit 3, and commands the identification unit 23 Identify the vanishing point of the image and the left and right lines. Step 820 is that the offset judging unit 3 judges whether there is a lane offset or a lane departure offset based on the received result, and closes step 821, otherwise 19! 27〇 827 ' to step 822. Step 821 is that the offset determining unit 3 sends the warning signal to the material, the display unit 5', and the warning unit 5 notifies the driver of the vehicle assisted by the warning signal 0 _ V 822 疋 the focus area detection The unit 22 receives the vanishing point recognized by the identification unit 23, and when the vanishing point is found and the concentration of the vanishing point is high enough, the key area is reduced to 1/3 of the original image size, and when the point is not found, the point is disappeared. The concentration of the point is too low, and the key area is expanded to 2/3 of the original image size J, and the key area is maintained at the remaining time, and the key area is transmitted to the image processing unit 21. Skip to step 8〇1. ▲ Step 823 is that the evaluation unit 25 outputs the currently received tracking result to the offset determining unit 3, and commands the tracking unit 24 to continue to track the vanishing point and the left and right lines of the next image. Step 824, the offset determination list & 3 determines whether there is a lane offset scale lane offset based on the received tracking result. If yes, off (four), no | then, go to step 809. In summary, the lane recognition system of the present invention and the lane having the system. (4) warning device and method thereof firstly make an edge image of the original image, and identify the vanishing point and the left and right lines in the edged image by the method of transducing (4) In addition, the key areas can be increased in time, so whether the vehicle is on a road with or without markings, or even other vehicles or debris, the lane can be identified, greatly improving the lane recognition, and The high lane recognition can avoid the decision of lane offset and greatly improve the accuracy of lane departure judgment. And when the vanishing point in the marginalized image and the left and right lines are recognized, 20
1270827 哈即可以速度較㈣料方式追訂—張邊緣化影像中的 ’失點及左、右邊線’並評估追縱結果是否合理，春不合 理時，再用韻方式㈣τ—張邊緣化影像巾的 左、右邊線’兩者搭配使用，不僅可以提高運算速度，又 不會降低車道的辨識度。 又 ▲惟以上所述者，僅為本發明之較佳實施例而已，當不 能以此限定本發明實施之範圍，即大凡依本發明中^利 範圍及發明說明内容所作之簡單的等效變化與修/皆仍 屬本發明專利涵蓋之範圍内。 【圖式簡單說明】 圖1是習知車道偏詩示裝置的方塊圖； 圖2疋一方塊圖，⑨明本發明車道辨識系統及里有續 系統的車道偏移警告裝置及其方法之較佳實施例；〃 圖3是一示意圖，說明該較佳實施例中消失點 覺水平線Μ的定義；及 、 圖4是該較佳實施例的流程圖。 21 1270827 【主要元件符號說明】 1 … • 視訊擷取單元 24— * 追縱單元 2… ♦ 車道辨識系統 25·… 評估單元 21 · · • 影像處理單元 3 偏移判斷單元 22· · ♦ 重點區域偵測單 5…* 警示單元 元 801〜824 步驟 23· · * 辨識單元1270827 Ha can speed up (four) material tracking - the 'missing point and left and right line' in the marginalized image and evaluate whether the tracking result is reasonable. When the spring is unreasonable, then use the rhyme method (4) τ - marginalized image The left and right lines of the towel are used together to improve the calculation speed without reducing the lane recognition. The above is only the preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, that is, the simple equivalent change according to the scope of the invention and the description of the invention. And repairs are still within the scope of the patent of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a conventional lane-shifting poem device; FIG. 2 is a block diagram showing a lane discriminating system of the present invention and a lane offset warning device and method therefor. BEST MODE FOR CARRYING OUT THE INVENTION Fig. 3 is a schematic view showing the definition of the horizontal point 消失 of the vanishing point in the preferred embodiment; and Fig. 4 is a flow chart of the preferred embodiment. 21 1270827 [Description of main component symbols] 1 ... • Video capture unit 24 - * Tracking unit 2... ♦ Lane recognition system 25... Evaluation unit 21 · · • Image processing unit 3 Offset judgment unit 22 · · ♦ Key area Detection list 5...* Warning unit 801~824 Step 23· · * Identification unit
22twenty two
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW94121884A TWI270827B (en) | 2005-06-29 | 2005-06-29 | Traffic lane recognition system and traffic lane deviation warning device with traffic lane recognition system and method of forming the same |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW94121884A TWI270827B (en) | 2005-06-29 | 2005-06-29 | Traffic lane recognition system and traffic lane deviation warning device with traffic lane recognition system and method of forming the same |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TW200701128A TW200701128A (en) | 2007-01-01 |
| TWI270827B true TWI270827B (en) | 2007-01-11 |
Family
ID=38430292
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW94121884A TWI270827B (en) | 2005-06-29 | 2005-06-29 | Traffic lane recognition system and traffic lane deviation warning device with traffic lane recognition system and method of forming the same |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TWI270827B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8687063B2 (en) | 2007-08-30 | 2014-04-01 | Industrial Technology Research Institute | Method for predicting lane line and lane departure warning system using the same |
| US8903638B2 (en) | 2011-08-26 | 2014-12-02 | Industrial Technology Research Institute | Warning method and system for detecting lane-changing condition of rear-approaching vehicles |
| US8902053B2 (en) | 2011-11-16 | 2014-12-02 | Industrial Technology Research Institute | Method and system for lane departure warning |
| US9412250B2 (en) | 2013-08-06 | 2016-08-09 | Nuvoton Technology Corporation | Alert method and alert system thereof |
| TWI676970B (en) * | 2017-04-20 | 2019-11-11 | 大陸商騰訊科技（深圳）有限公司 | Traffic lane selecting method, target vehicles, and computer storage medium |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI587220B (en) * | 2011-08-26 | 2017-06-11 | Maction Technologies Inc | Lane Identification System with Hoggson Conversion Principle and Its |
-
2005
- 2005-06-29 TW TW94121884A patent/TWI270827B/en not_active IP Right Cessation
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8687063B2 (en) | 2007-08-30 | 2014-04-01 | Industrial Technology Research Institute | Method for predicting lane line and lane departure warning system using the same |
| US8903638B2 (en) | 2011-08-26 | 2014-12-02 | Industrial Technology Research Institute | Warning method and system for detecting lane-changing condition of rear-approaching vehicles |
| US8902053B2 (en) | 2011-11-16 | 2014-12-02 | Industrial Technology Research Institute | Method and system for lane departure warning |
| US9412250B2 (en) | 2013-08-06 | 2016-08-09 | Nuvoton Technology Corporation | Alert method and alert system thereof |
| TWI587249B (en) * | 2013-08-06 | 2017-06-11 | 新唐科技股份有限公司 | Driving alert method and alert system thereof |
| TWI676970B (en) * | 2017-04-20 | 2019-11-11 | 大陸商騰訊科技（深圳）有限公司 | Traffic lane selecting method, target vehicles, and computer storage medium |
| US11059485B2 (en) | 2017-04-20 | 2021-07-13 | Tencent Technology (Shenzhen) Company Limited | Lane selection method, target vehicle and computer storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| TW200701128A (en) | 2007-01-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TWI270827B (en) | Traffic lane recognition system and traffic lane deviation warning device with traffic lane recognition system and method of forming the same | |
| WO2022078463A1 (en) | Vehicle-based obstacle detection method and device | |
| CN102765365B (en) | Pedestrian detection method based on machine vision and pedestrian anti-collision warning system based on machine vision | |
| Low et al. | Simple robust road lane detection algorithm | |
| CN109649384A (en) | A kind of parking assistance method | |
| CN108009475A (en) | Driving behavior analysis method, apparatus, computer-readable recording medium and electronic equipment | |
| CN105678787A (en) | Heavy-duty lorry driving barrier detection and tracking method based on binocular fisheye camera | |
| CN103679171B (en) | A gait feature extraction method based on human body gravity center track analysis | |
| Kuk et al. | Fast lane detection & tracking based on Hough transform with reduced memory requirement | |
| TW201225004A (en) | Image based detecting system and method for traffic parameters and computer program product thereof | |
| TW201226234A (en) | Image type obstacle detection car reversing warning system and method | |
| KR101663574B1 (en) | Method and system for detection of sudden pedestrian crossing for safe driving during night time | |
| CN105182320A (en) | Depth measurement-based vehicle distance detection method | |
| JP2013008328A (en) | Vehicle periphery monitoring system | |
| JP4940177B2 (en) | Traffic flow measuring device | |
| JP2000285245A (en) | Method and device for preventing collision of moving body and recording medium | |
| CN107765246A (en) | A kind of effective target recognition methods based on the detection of millimeter wave dual radars | |
| CN102768726A (en) | Pedestrian detection method for preventing pedestrian collision | |
| CN104200199A (en) | TOF (Time of Flight) camera based bad driving behavior detection method | |
| US9076034B2 (en) | Object localization using vertical symmetry | |
| Bhowmick et al. | Stereo vision based pedestrians detection and distance measurement for automotive application | |
| Sehestedt et al. | Efficient lane detection and tracking in urban environments | |
| WO2021022643A1 (en) | Method and apparatus for detecting and tracking target in videos | |
| CN108627135A (en) | A kind of method and system for the road conditions that are recessed by front | |
| TW201520907A (en) | Vision-based cyclist and pedestrian detection system and the methods |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MM4A | Annulment or lapse of patent due to non-payment of fees |