201144108 六、發明說明： 【發明所屬之技術領域】 本發明係關於測定架線及轨道等之保養所使用之電氣化 鐵路保養用車輛之行駛位置之電氣化鐵路保養用車輛位置 測定裝置，尤其係關於利用圖像處理取得電氣化鐵路保養 用車輛之行駛位置與通過時刻之關係之電氣化鐵路保養用 車輛位置測定裝置。 【先前技術】 作為電氣化鐵路之設備，主要可舉出架線與軌道。該等 分別為鐵路營運之重要保養設備。架線在運用電氣化鐵路 車輛中每當電氣化鐵路車輛通過時會與集電裝置接觸。因 此架線逐漸磨耗，若未作更換將有最終因斷裂而招致事故 之虞。又’軌道會因電車之運行而產生彎曲、庫變、損 耗等1放任不管該等之進展，則有招致電氣化鐵 路車輛之脫軌等事故之虞。 為迅速掌握因架線或軌道之磨耗等而需要注意處 (=應广道，特定架線或軌道之需要注意處之位置 ^ =某地點行駛幾公里之位置存在需要注意處等）之資 λ，由保養官理之面而言為重要事項。 養線或執道等設備之保養之情财，使用保 意處之情形時，由取得其資料時之車 斷 處之位置（例如參照下述專利文獻】）。置推斷需要庄意 此處，作為衫車辆位置之方法，先前例如有使用自動 J55389.doc 201144108 列車控制裝置（ATC)或旋轉編碼器之方法，分別具有以下 特徵》 、 自動列車控制裝置係控制列車之位置或速度者，原為用 以運轉車輛之系統，但該自動列車控制褒置中設有將列車 之位置或速度之資訊向外部通信之機制。因此，使用自動 列車控制裝置特定車輛位置之情形，係利用與該自動列車 控制裝置之外部進行通信之機制而特定保養專用車輛之位 置（例如參照下述專利文獻2)。 又，使用旋轉編碼器特定車輛位置之情形，係基於由旋 轉編碼器而得之車輪之迴轉數，由該車輪之迴轉數與車輪 之圓周之長度而求得以某地點為基準位置之車輛之行駛距 離（例如參照下述專利文獻3)。 [先前技術文獻] [專利文獻] [專利文獻1]日本特開2008-89523號公報 [專利文獻2]曰本特開2006-248412號公報 [專利文獻3]日本特開2004-17700號公報 [非專利文獻] [非專利文獻1]藤吉弘亘「斜率基底之特徵提取_SIFT與 HOG-」資訊處理學會研究報告CVIM 160、2007年、 P-211-224 【發明内容】 [發明所欲解決之問題] 但’利用上述先前之自動列車控制裝置之情形，不僅需 155389.doc 201144108 要在車輛側設置收發機器’亦需要在軌道側以一定間隔設 置收發裝置’因此設置於長距離之線區之情形中，有個數 增多、耗費成本之問題。 又’由於旋轉編碼器為藉由計算車輪之迴轉角而測定行 駛距離之構成，因此有以下問題··因刹車或坡道行駛等使 得列車雖未行進但車輪仍在迴轉，或因車輪打滑使得車輪 雖未迴轉但列車仍在行進等局部性產生之誤差累加，導致 需要注意處之位置特定變困難之虞。 因此，先前之自動列車控制裝置中，於車輛之車頂上設 置線感測器相機，將利用該線感測器（以下稱作線感測器） 拍攝之架線附近之圖像進行圖像處理（模版比對或邊緣檢 測等）而檢測吊架，基於拍攝該吊架之時刻而特定車輛之 行駛位置與通過時刻之關係，藉此消除上述誤差。 但，若吊架自身傾斜伸出或吊架傾斜映照於畫面上，則 利用模版比對進行吊架檢測之情形中會有與模版之對照時 吊架檢查之精度低落之虞，而利用邊緣檢測進行吊架之檢 測之情形中，有因縱向邊緣弱化而使得吊架檢測之精度低 落之虞。 由上，本發明之目的係提供一種可高精度地特定車輛之 行駛位置與通過時刻之關係之電氣化鐵路保養用車輛位置 測定裝置。 [解決問題之技術手段] 用以解決上述問題之第丨發明之電氣化鐵路保養用車輛 位置測定裝置之特徵在於具備： 155389.doc 201144108 線感測器’其拍攝架線或懸吊支持架線 之圖像； 〜p木踝 圖像建立部’其將前述線錢^❹攝之 版圖像與輪入圖像； 馮模 記憶部’其保存在前述圖像建 像與前述輸人圖像； =立處理部，其從保存於前述記憶部之前述模版圖 2取了特徵量，建立模版圖像資料並保存於前述記憶 耐特徵量提取處理部，其從保存於前述記憶部之前述 輸入圖像提取SIFT特徵量並保存於前述記憶部内； 予架位置檢測處理部，其探Μ存料述記憶部之前述 模版資料之耐特徵4與前述輸人輯之贿特徵量之對 應點，而檢測前述輸人圖像中之吊架位置並保存 憶部内》 用以解決上述問題之第2發明之電氣化鐵路保養用車輛 位置測定裝置之特徵在於，前述第i發明之電氣化鐵路保 養用車輛位置測定裝置中， 具備時刻記錄處理部，其基於前述線感測器之抽樣頻率 與保存於前述記㈣之前述輸人圖像上之吊架位置，計算 前述吊架經檢測之時刻，並將時刻資料保存於前述記憶部 内0 用以解決上述問題之第3發明之電氣化鐵路保養用車輛 位置測定裝置之特徵在於：前述第1發明或第2發明之電氣 155389.doc 201144108 化鐵路保養用車輛位置測定裝置中， 於前述吊架位置檢測處理部中，當存在上次之吊架位置 之情形時，交遞上次之吊牟 巾朱位置與吊架線及架線之傾斜資 料，進行以上次之吊架位置為基準而限定處理區域之次處 理區域之計算。 蛟 用以解決上述問題之笛4久 鸿之弟4發明之電氣化鐵路保養用車 位置測定裝置之#I & 击紅 置之特徵在於，第3發明之電氣化鐵路保養用 車輛位置特定裝置中， 於引述吊木位置檢測處理部中’擴大前述次處理區域。 [發明之效果] 根據本發明，可接视 種可高精度地特定車輛之行駛位 置了1過時刻之關係之電氣化鐵路保養用車輛位置測定裝 【實施方式】 乂: #對本發明之電氣化鐵路保養用車輛位置測定裝 置之實施例，參照附圖進行說明。 、 [實施例1] 夕 十子本發明之電氣化鐵路保養用車輛位置測定裝 置之第1實施例進行說明。 、 ^實施例之電氣化鐵路保養用車輛位置測定裝置之特徵 向枕太Γ用叹於車輛之車頂上之線感測器，使線感測器朝 二肖、且朝向斜上方拍攝吊架線與架線，求得 大小或迴轉變化不易 變特忾― 4曰之圖像特徵量即SIFT(尺度不 ^将徵轉換）特徵量（炎 (> “，、上述非專利文獻υ ,求得與從預 I55389.doc 201144108 之對應點，藉此 先求得之模版用之圖像提取之SIFT特徵量 檢測吊架。 首先，針對本實施例之電氣化鐵路保養用車輛位置測定 裝置之概要進行說明。 圖1係顯示本實施例之電氣化鐵路保養用車輛位置測定 裝置之概要之模式圖。 如圖1所示，本實施例之電氣化鐵路保養用車輛位置測 定裝置具備：線感測m，其係為了拍攝架線1()1或懸吊支 持架線HH之吊架102之吊架線103之圖像，而於車輛⑽之 車頂上朝向枕木方向、且朝向斜上方設置，於圖】中箭頭L 所示之攝影線上拍攝圖像；照明2,其為了在夜間或随道 内亦可利用線感測器1拍攝圖像，而設於線感測器丄之周 邊；-處理用個人電腦3,其保存線感測器⑽拍攝之圖像並 進行圖像處理，且設於車輛1()()内。並且，㈣測器與處 理用個人電腦3以可收發圖像資料之方式連接。另，圖！中 箭頭A所不之方向表示車輛1〇〇之行進方向。 接者’針對本實施例之電氣化鐵路保養用車輛位置測定 裝置之構成進行說明。 圖2係顯示本實施例之電氣化鐵路保養用車輛位置測定 裝置之構成之方塊圖。 如圖2所不’本實施例之電氣化鐵路保養用車辅位置測 定裝置具備：圖像建立部1〇，其將線感測^所拍攝之圖 像，立為模版圖像與輸入圖像·記憶體η，其保存圖像建 立。Ρ ίο中建立之模版圖像與輸入圖像；模版建立處理部 155389.doc 201144108 13，其根據保存於記憶體u之模版圖像提取811^特徵量， 建立模版資料並保存於記憶體丨2内；SIFT特徵量提取處理 部丨4 ’其從保存於記憶體u之輸入圖像提取811?丁特徵量並 保存於記憶體12内；予架位置檢測處理部15，其探索保存 於記憶體12之模版資料之SIFT特徵量與輸入圖像之sift特 徵量之對應點，檢測圖像中之吊架位置並保存於記憶體12 内；時刻記錄處理部16，其基於線感測器丨之抽樣頻率與 保存於記憶體12之圖像上之吊架位置，算出檢測吊架ι〇2 之時刻’並將時刻資料保存於記憶體12内。 另，本實施例中，處理用個人電腦3發揮圖像建立部 10、記憶體11、12、模版建立處理部13、SIFT特徵量提取 處理部14、吊架位置檢測處理部15、爷架位置檢測處理部 b及時刻記錄處理部16之作用。又，記憶體丨丨與記憶體^ 亦可為共通之記憶體。 接著，針對本實施例之電氣化鐵路保養用車輛位置測定 裝置之處理順序進行說明。 圖3係顯示本實施例之電氣化鐵路保養用車輛位置測定 裝置之處理順序之流程圖。 如圖3所示，作為本實施例之電氣化鐵路保養用車輛位 置測定裝置之處理順序’首先於步驟ρι〇中，從線感測器工 取得抽樣用圖像。$，如圖4所示，模版圖像為切割出所 拍攝之1個吊架1 〇2之部份之圖像。 接著，於步驟P11 建立模版資料。另， t，從模版圖像中提取81?丁特徵量而 如圖5中黑色圓點所示，81打特徵量 155389.doc 201144108 於模版圖像上存在複數個，建立其所有的模版資料。另， 對於SIFT特徵量之求取方法m述料利文獻⑽ 記載之方法，此處省略說明。 接著，於步驟？12中，從線感測器i建立輸入圖像。另， 如圖6所示，輸入圖像之縱軸為高度，橫軸為時間。 接著，於步驟P13中，從輸入圖像提取SIFT特徵量。 接著，於步驟P14中，進行模版資料之SIFT特徵量與輸 入圖像之SIFT特徵量之對應點探索’檢測輸入圖像内之所 有吊架位置…吊架位置之檢測方法係計算模版資料之 SIFT特徵量與輸入圖像之所有SIFT特徵量之歐幾里德距 離’將最小值者作為對應點。又，同樣對模版資料之:有 SIFT特徵量求取歐幾里德距離，基於各個對應點而 架位置。 此處’所謂歐幾里德距離，係將模版資料之㈣特徵量 與輸入圖像之贿特徵量之差平方，計算出該等之總和平 方根者。具體言之，隨特徵量為128次元之特徵量。因 此，歐幾里德距離之計算如下式（1)。 [數1][Technical Field] The present invention relates to an electric railway maintenance vehicle position measuring device for measuring the traveling position of an electrified railway maintenance vehicle used for maintenance of a wire, a rail, or the like, and more particularly to the use map. An electric vehicle maintenance vehicle position measuring device that processes the traveling position of the electrified railway maintenance vehicle and the passing time. [Prior Art] As an equipment for an electrified railway, a wire and a rail are mainly exemplified. These are important maintenance equipment for railway operations. The wiring is in contact with the current collecting device whenever the electrified railway vehicle passes through the use of the electrified railway vehicle. As a result, the wiring is gradually worn out, and if it is not replaced, there will be an accident caused by the break. In addition, the track will be bent, stored, and damaged due to the operation of the train. 1 Regardless of the progress, there will be accidents such as derailment of electrified railway vehicles. In order to quickly grasp the wear and tear of the wire or the track, etc. (= should be Guangdao, the location of the specific line or track needs to pay attention to the location ^ = a few kilometers at a certain location, there is a need to pay attention to, etc.) It is an important matter in terms of maintenance. In the case of the maintenance of equipment such as the maintenance of the line or the road, the position of the vehicle at the time of obtaining the information (for example, refer to the following patent documents). Inference is required here, as a method of the position of the shirt vehicle, previously, for example, the method using automatic J55389.doc 201144108 train control device (ATC) or rotary encoder has the following features respectively, automatic train control device control The position or speed of the train is originally a system for operating the vehicle, but the automatic train control unit is provided with a mechanism for communicating the information of the position or speed of the train to the outside. Therefore, in the case where the automatic train control device is used to specify the vehicle position, the position of the maintenance-dedicated vehicle is specified by a mechanism for communicating with the outside of the automatic train control device (see, for example, Patent Document 2 below). Moreover, the case where the rotary encoder is used to specify the position of the vehicle is based on the number of revolutions of the wheel obtained by the rotary encoder, and the number of revolutions of the wheel and the length of the circumference of the wheel are used to obtain the vehicle with the position as the reference position. The distance (for example, refer to Patent Document 3 below). [PRIOR ART DOCUMENT] [Patent Document 1] JP-A-2008-89523 [Patent Document 2] JP-A-2006-248412 (Patent Document 3) JP-A-2004-17700 Non-Patent Document] [Non-Patent Document 1] Fujiji Hiroshi, "Characteristic Extraction of Slope Base_SIFT and HOG-" Information Processing Society Research Report CVIM 160, 2007, P-211-224 [Summary of the Invention] [Invented The problem] But in the case of using the above-mentioned automatic train control device, it is necessary to set not only the 155389.doc 201144108 to set up the transceiver on the vehicle side but also to set the transceiver at a certain interval on the track side. In the case of the situation, there is a problem of an increase in the number and cost. In addition, since the rotary encoder is configured to measure the travel distance by calculating the swing angle of the wheel, there are the following problems: • The train is not traveling but the wheel is still turning due to braking or ramp running, or the wheel is slipping Although the wheels are not turned but the train is still traveling, the errors caused by the locality are accumulated, which makes it difficult to pay attention to the position of the position. Therefore, in the conventional automatic train control device, a line sensor camera is disposed on the roof of the vehicle, and an image near the overhead line photographed by the line sensor (hereinafter referred to as a line sensor) is used for image processing ( The stencil is detected by stencil matching or edge detection, and the relationship between the traveling position of the specific vehicle and the passing time is determined based on the time at which the cradle is photographed, thereby eliminating the above error. However, if the hanger itself is tilted out or the tilt of the hanger is reflected on the screen, the accuracy of the hanger inspection when the hanger is detected by the template comparison may be lowered, and the edge detection is utilized. In the case of the detection of the hanger, the accuracy of the hanger detection is lowered due to the weakening of the longitudinal edge. In view of the above, an object of the present invention is to provide an electric railway maintenance vehicle position measuring device which can precisely specify the relationship between the traveling position of the vehicle and the passing time. [Technical means for solving the problem] The vehicle position measuring device for electrified railway maintenance according to the third invention for solving the above problems is characterized by: 155389.doc 201144108 Line sensor' image of the shooting line or the hanging support line ;p p 踝 建立 建立 建立 ' ' ' ' ' ' ' ' ' ' ' ' ' ' 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯 冯The processing unit obtains a feature amount from the template 2 stored in the memory unit, creates template image data, and stores the template image data in the memory-resistance feature amount extraction processing unit, and extracts the image from the input image stored in the memory unit. The SIFT feature quantity is stored in the memory unit; the pre-position detection processing unit detects the corresponding point of the resistance characteristic 4 of the template data of the memory unit and the bribe feature quantity of the input series, and detects the loss. The electric vehicle maintenance vehicle position measuring device according to the second invention for solving the above-mentioned problems is characterized in that the electric power of the first invention The vehicle position measuring device for railway maintenance includes a time recording processing unit that calculates the hanger by the sampling frequency of the line sensor and the hanger position stored in the input image of the above (4) At the time of the present invention, the electric vehicle maintenance vehicle position measuring device according to the third invention for solving the above-mentioned problems is characterized in that the first invention or the second invention is electrically 155389.doc 201144108 In the railway maintenance vehicle position measuring device, when the last hanger position is present in the hanger position detecting processing unit, the tilting information of the last hanging towel and the hanger wire and the wire are transferred. The calculation of the secondary processing area of the processing area is defined based on the above-mentioned hanger position as a reference. # I 4 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 久 电气 电气 电气In the reference to the hanging position detection processing unit, the secondary processing area is expanded. [Effect of the Invention] According to the present invention, an electric railway maintenance vehicle position measuring device capable of accurately specifying the relationship between the traveling position of the vehicle and the time of the vehicle can be viewed in a high-performance manner. [Embodiment: #: The electrified railway maintenance of the present invention An embodiment of the vehicle position measuring device will be described with reference to the drawings. [Embodiment 1] The first embodiment of the vehicle position measuring device for electrified railway maintenance according to the present invention will be described. ^The characteristics of the vehicle position measuring device for electrified railway maintenance in the embodiment are used to sigh the line sensor on the roof of the vehicle, so that the line sensor is directed toward the second and obliquely upward to shoot the hanger wire and the wire. It is not easy to change the size or the change of the rotation. The image feature quantity of the image is the SIFT (the scale is not converted) feature (inflammation (> In the corresponding point of I55389.doc 201144108, the SIFT feature quantity detection hanger for image extraction for the template is first obtained. First, the outline of the vehicle position measuring device for electrified railway maintenance according to the present embodiment will be described. A schematic diagram showing an outline of the vehicle position measuring device for electrified railway maintenance according to the present embodiment. As shown in Fig. 1, the vehicle position measuring device for electrified railway maintenance of the present embodiment includes a line sensing m for taking a line. 1()1 or the image of the hanger line 103 of the hanger 102 supporting the support line HH, and placed on the roof of the vehicle (10) toward the sleeper and facing obliquely upward, as shown by the arrow L in the figure Online shooting image; illumination 2, which can be used to capture images by line sensor 1 at night or in the track, and is disposed around the line sensor ;; - processing personal computer 3, its save line sensing The image captured by the device (10) is image-processed and disposed in the vehicle 1()(). Moreover, the (four) detector and the processing personal computer 3 are connected in such a manner as to transmit and receive image data. The direction indicated by A indicates the traveling direction of the vehicle 1. The following is a description of the configuration of the vehicle position measuring device for electrified railway maintenance according to the present embodiment. Fig. 2 shows the position measurement of the electric railway maintenance vehicle of the present embodiment. A block diagram of the configuration of the device. As shown in Fig. 2, the auxiliary position measuring device for electrified railway maintenance of the present embodiment includes an image creating unit 1 that sets the image captured by the line sensing into a template. Image and input image·memory η, which saves image creation. 模 ίο created template image and input image; template creation processing unit 155389.doc 201144108 13, according to the template saved in memory u Like extracting 811^ features The template data is created and stored in the memory 丨2; the SIFT feature quantity extraction processing unit 丨4' extracts 811 metric feature values from the input image stored in the memory u and stores them in the memory 12; The position detection processing unit 15 searches for a corresponding point of the SIFT feature amount of the template data stored in the memory 12 and the sift feature amount of the input image, detects the hanger position in the image, and stores it in the memory 12; The processing unit 16 calculates the time at which the hanger ι 2 is detected based on the sampling frequency of the line sensor 与 and the hanger position stored on the image of the memory 12, and stores the time data in the memory 12. In the present embodiment, the processing personal computer 3 functions as the image creating unit 10, the memory 11, 12, the template establishing processing unit 13, the SIFT feature amount extracting processing unit 14, the hanger position detecting processing unit 15, and the position of the rack. The functions of the detection processing unit b and the time recording processing unit 16 are as follows. Moreover, the memory 丨丨 and the memory ^ can also be a common memory. Next, the processing procedure of the vehicle position measuring device for electrified railway maintenance according to the present embodiment will be described. Fig. 3 is a flow chart showing the processing procedure of the vehicle position measuring device for electrified railway maintenance according to the present embodiment. As shown in Fig. 3, the processing sequence of the vehicle position measuring device for electrified railway maintenance according to the present embodiment is first obtained in step ρι〇 from the line sensor. $, as shown in Fig. 4, the template image is an image of a portion of the hanger 1 〇 2 taken. Next, template data is created in step P11. In addition, t, extracting the 81-square feature from the template image and as shown by the black dot in Fig. 5, the 81-characteristic amount 155389.doc 201144108 has a plurality of templates on the template image to establish all of the template data. Further, the method for obtaining the SIFT feature amount is described in the document (10), and the description thereof is omitted here. Then, in the steps? In 12, an input image is created from the line sensor i. In addition, as shown in FIG. 6, the vertical axis of the input image is height, and the horizontal axis is time. Next, in step P13, the SIFT feature amount is extracted from the input image. Next, in step P14, the corresponding point of the SIFT feature quantity of the template data and the SIFT feature quantity of the input image is searched for 'detecting all the hanger positions in the input image. The detection method of the hanger position is the SIFT of calculating the template data. The Euclidean distance of the feature quantity and all SIFT feature quantities of the input image is taken as the corresponding point. Also, for the template data: the SIFT feature quantity is used to obtain the Euclidean distance, and the position is based on each corresponding point. Here, the so-called Euclidean distance is obtained by squaring the difference between the (4) feature quantity of the template material and the bribe feature quantity of the input image to calculate the total square root of the sum. Specifically, the feature quantity is a feature quantity of 128-dimensional. Therefore, the Euclidean distance is calculated as shown in the following equation (1). [Number 1]
上述式Ο)中，v意指輸入圖像之SIFT特徵量，yt意指模 版資料之SIFT特徵量，d意指歐幾里德距離。 " 另’通常’歐幾里德距離成為最小之處在i張輸入圖像 155389.doc 201144108 内八s出現1處。又，根據車輛100之行駛速度不同，在】 張輸入圖像内可能會拍攝到複數之吊架102，如此一來將 無法檢測所有吊架丨〇2。 仁由線感測器1拍攝之橫軸為時間之輸入圖像t，於 同一時間上不存在複數之吊架1〇2。因此，本實施例中， #圖7中虛線所# ’將i張輸人圖像依每小時劃分成短條狀 而進订對應點探索，因而可檢測複數之吊架1〇2。另，劃 分之寬度為任意，但基本上與模版資料之寬度為相同程 度。 接者，於步驟P15中，從所檢測之吊架位置計算時刻並 記錄時刻。如圖6所示，輸入圖像之橫軸為時間。然後， 基於線感測器丨之抽樣頻率與輸入圖像上之吊架位置 算出時刻》 例如，線感測器1之抽樣頻率為！ kHz，在時間軸方向之 第100條線上檢測吊架102之情形中，每W之攝影時間為 l[sec卜1000[Hz]=0.001[sec] ’而在第1〇〇條線上檢測吊架 102，故為 [線]=〇 。 最後，於步驟P16中，重複步驟P12〜16直至對所有的輸 入圖像之處理結束為止，在對所有的輸入圖像之處理結束 之時間點，結束本實施例之電氣化鐵路保養用車輛位置測 定裝置之一連串處理。 如上說明，根據本實施例之電氣化鐵路保養用車輛位置 測定裝置，藉由進行使用模版資料之811?1特徵量與輸入圖 像之SIFT特徵量之對應點探索處理，而如圖6所示，即使 155389.doc 201144108 吊架102與周圍之構造物1〇4存在於同一處之情形令亦可 檢測爷架1G2，因此可防止吊架IG2之漏檢測。 又，根據本實施例之電氣化鐵路保養用車輛位置測定裝 置由於使用對於比例、迴轉或明亮度變化具穩健性之 特徵量因此無需對應於比例、迴轉或明亮度變化而 準備所有情況之模版資料，即使在比例、迴轉或明亮度有 戶:變化之情形令，只要有1個模版資料便可檢測，因此可 提高吊架102之檢測效率。 [實施例2] 以下，針對本發明之電氣化鐵路保養用車輛位置測定裝 置之第2實施例進行說明。 本實施例之電氣化鐵路保養用車輛位置測定裝置係與第 貫施例之電氣化鐵路保養用車輛位置測定裝置大致相同 之構成，但具有以下不同點：當存在上次之吊架位置之情 形時，以上次之吊架位置為基準而限定處理區域，藉此謀 求處理之高速化。 接著，針對本實施例之電氣化鐵路保養用車輛位置測定 裝置之構成進行說明。 圖8係顯示本實施例之電氣化鐵路保養用車輛位置測定 裝置之構成之方塊圖。 如圖8所示，本實施例之電氣化鐵路保養用車輛位置測 定裝置之構成與第丨實施例之電氣化鐵路保養用車輛位置 測疋裝置之構成大致相同，但具有以下不同點：於吊架位 置檢測處理部1 5中，存在上次之吊架檢測位置之情形時， 155389.doc 201144108 交遞上次之吊架位 進行以上次之吊架 域之計算。 置與吊架線103及架線1〇1之傾斜資料， 位置為基準而限定處理區域之次處理區 =，針對實施例之電氣化鐵路保養用車辅位置測定裝 置之處理順序進行說明。 本實施例之電氣化鐵路保養用車輛 順序與第1實施狀電氣㈣耗翻車心置 之構成大致_’但具有以下不同點：當存在上次增 位置之情形時，追加以上次之吊架位置為基準而限定處理 區域之次處理區域之計算處理。 SIFT特徵量由於在算出處理上過於花費時間，因此需要 謀，處理之高速化。SIFF特徵量之算出所需要之時間與處 理區域之大小成比例’因此若可Μ處理區域則可有效率 ，進行處理。此處’由於吊架102必定存在於吊架線如與 架線1〇1之間，且存在於吊架線1()3與架線1G1之傾斜方向 之可能性高’因此若可檢測第1個吊架1〇2,則可限定處理 ，域。如此，由於可根據吊架線103及架線1〇1之傾斜而限 定次處理區域，因此可謀求處理之高速化。 圖9係顯示本實施例t電氣化鐵路保養用|輛位置測定 裝置之處理順序之流程圖。 如圖9所示，作為本實施例之電氣化鐵路保養用車輛位 置測定裝置之處理順序’首先於步驟？1()中，從線感測器^ 取知模版用SI像。另’如圖4所示’模版圖像為切割出所 拍攝之1個吊架102之部份之圖像。 I55389.doc -13· 201144108 接著，於步驟Pllt，從模版圖像提取SIFT特徵量而建 立模版資料。3，如圖5中黑色圓點所*，SIFT特徵量於 模版圖像上存在複數個，建立其所有之模版資料。另，對 於SIFT特徵量之求取方法，於上述非專利文獻1中有記 載’因此此處省略說明。 接考，於步驟P12中，從線感測器i建立輸入圖像。另， 如圖6所示，輸入圖像之縱軸為高度，橫軸為時間。 接著，於步驟P20中，若存在上次之吊架位置之情形 時，於步驟P2I中執行次處理區域之計算處理。又不存 在上-人之爷架位置之情形時，於步驟pi3中根據輸入圖像 全體進行吊架位置檢測處理。 接著，於步驟P21中，從第j張圖像中檢測吊架ι〇2，基 於第1張圖像之吊架檢測位置4及吊架1〇3與架線⑻之傾斜 而決定次處理區域5。具體言之，如圖1〇所示，吊架線Μ] 與架線ιοί之傾斜為右下方向之情形時，吊架1〇2存在於吊 架線103與架線1〇1之間，因此存在於吊架線ι〇3與架線"I 之傾斜方向之可能性較高。因此，可以第丨張圖像…4 為基準’在吊㈣H)3與架線1G1之傾斜方向決定次處理區 域5。 另，將圖10中A、B所示之一點鏈線作為圓像之切售 線，將1點鏈線A左側作為第丨張圖像，將—點鏈線a與γ 點鏈線B之間作為第2張圖像，將一點鏈線8右側作為、笛 張圖像。 ·’、、弟 接著，於步驟IM4中’在次處理區域5内或圆像全體進七 I55389.doc 201144108 吊架位置檢測處理。 :二於步驟P15中’從檢測出之吊架位置計算時刻並 3己録時刻β 最後:於步驟Ρ16中，若存在上次之吊架位置之情形 日-’重硬步驟Ρ12〜15直至輪人圖像不存在為止，在對 的輸入圖像結束處理之時 .. 4.，、 ，、，'°束本實施例之電氣化鐵 保養用車辆位置測定裝置之-連串處理。 測定月’根據本實施例之電氣化鐵路保養用車輛位置 疋、不僅發揮第丨實施例之電氣化鐵路 位置測定裝詈之对s _ 货用早桶 ， 效果，且糟由將處理區域限定於區域5， 亦可謀求處理之高速化。 一 [實施例3] 番以Γ 一針對本發明之電氣化鐵路保養用車輛位置測定裝 置之第3實施例進行說明。 、 本實施例之電氣化鐵路保養用車輛位置 貫施例之電氣化鐵路 1/、弟2 M . , 保養用車_位置測定裝置大致為相同 構成’但具有以下不同 _ ^ B + 將處理區域限定於次處理區域 度時’藉由擴大次處理區域5而提高吊架位置檢測精 穿=虚：對本實施例之電氣化鐵路保養用車輛位置測定 裝置之處理順序進行說明。 將圖11中A所示之—戥 點鏈線作為圖像之切割線，將一點 ，· .·衣之左側作為第丨張圖 2張圖像。如圖U所示 二點鏈線A之右側作為第 於圖像之切割線上拍攝到吊架1〇2 155389.doc -15- 201144108 之情形時，SIFT特徵量會有所不同’因此可能無法檢測出 吊架102。因此，如圖U所示，藉由將次處理區域5之範圍 擴大為包含前圖像、即包含第丨張圖像之一部份，而可檢 測在圖像之切割線上拍攝到之吊架丨〇2。 如上說明，根據本實施例之電氣化鐵路保養用車輛位置 測疋裝置’除了發揮第2實施例之電氣化鐵路保養用車輛 位置測疋裝置之效果外，並藉由擴大次處理區域5，如圖 11所示，即使在圖像之切割線上拍攝吊架1〇2之情形下， 亦可高精度地檢測吊架102。 [產業上之可利用性] 本發明係適用於架線及軌道等之保養所使用之側定電氣 鐵路車輛之位置之電氣化鐵路保養用車輛位置測定裝置 【圖式簡單說明】 1係顯示本發明之第i實施例之電氣化鐵路保養用車輛 置測定裝置之構成之模式圖 2係顯示本發明之第i實施例之電氣化鐵路 置測定裝置之構成之方_ 用皁輛 圖3係顯示本發明之第丨實施例之電氣化鐵路保 位置測定裝置之處理順序之流程圖。 圖4係顯示本發明之第丨實施例之電氣化鐵路保養 位置測定裝置之模版圖像之例之圖。 圖5係顯示本發明之第1實施例之電氣化鐵路保養 位置測定裝置之模版資料之SIFT特徵量之例之圖。 養用車輛 用車輛 用車輛 155389.doc •16- 201144108 圖6係顯示本發明之第1實施例之電氣化鐵路保養用車輛 位置測定裝置之輸入圖像之例之圖。 圖7係顯示本發明之第1實施例之電氣化鐵路保養用車輛 位置測定裝置之3吓丁特徵量之對應點探索處理區域之例之 圖。 圖8係顯示本發明之第2實施例之電氣化鐵路保養用車輛 位置測定裝置之構成之方塊圖。 圖9係顯示本發明之第2實施例之電氣化鐵路保養用車輛 位置測定裝置之處理順序之流程圖。 圖1 〇係顯示本發明之第2實施例之電氣化鐵路保養用車 輛位置測定裝置之次處理區域之例之圖。 圖11係顯示將本發明之第3實施例之電氣化鐵路保養用 車輛位置測定裝置之次處理區域放大之例之圖。 【主要元件符號說明】 1 線感測器 2 照明 3 處理用個人電腦 4 吊架位置 5 次處理區域 10 圖像建立部 11、12 記憶體 13 模版建立處理部 14 SIFT特徵量提取處理部 15 吊架位置檢測處理部 155389.doc -17- 201144108 16 時刻記錄處理部 100 車輛 101 架綠 102 吊架 103 吊架線 104 結構物 155389.doc -18-In the above formula ,), v means the SIFT feature quantity of the input image, yt means the SIFT feature quantity of the template data, and d means the Euclidean distance. " Another 'usually' Euclidean distance becomes the smallest place in the input image of 155389.doc 201144108. Further, depending on the traveling speed of the vehicle 100, a plurality of hangers 102 may be photographed in the input image, so that all the hangers 2 cannot be detected. The horizontal axis of the line sensor 1 is the input image t of time, and there is no plurality of hangers 1〇2 at the same time. Therefore, in the present embodiment, the dotted line #' in Fig. 7 divides the i-input image into short strips every hour and subscribes to the corresponding point search, so that the plurality of hangers 1〇2 can be detected. In addition, the width of the division is arbitrary, but is substantially the same as the width of the template data. Next, in step P15, the time is calculated from the detected hanger position and the time is recorded. As shown in Fig. 6, the horizontal axis of the input image is time. Then, based on the sampling frequency of the line sensor and the position of the hanger on the input image, the time is calculated. For example, the sampling frequency of the line sensor 1 is! kHz, in the case of detecting the hanger 102 on the 100th line in the time axis direction, the shooting time per W is 1 [sec 120 [Hz] = 0.001 [sec] ' and the hanger is detected on the 1st line 102, so [line] = 〇. Finally, in step P16, steps P12 to 16 are repeated until the processing of all the input images is completed, and the position measurement of the electric railway maintenance vehicle of the present embodiment is ended at the time when the processing of all the input images is completed. One of the devices is processed in series. As described above, the vehicle position measuring apparatus for electrified railway maintenance according to the present embodiment performs the search processing by using the corresponding point of the 811?1 feature amount of the template data and the SIFT feature amount of the input image, as shown in FIG. Even if the hanger 102 is present at the same place as the surrounding structure 1〇4, the rack 1G2 can be detected, so that the leak detection of the hanger IG2 can be prevented. Further, the vehicle position measuring device for electrified railway maintenance according to the present embodiment does not need to prepare template data for all situations corresponding to changes in scale, rotation, or brightness, since the feature amount is robust to the change in ratio, rotation, or brightness. Even if there is a ratio of scale, rotation or brightness: the situation of change, as long as one template data can be detected, the detection efficiency of the hanger 102 can be improved. [Embodiment 2] Hereinafter, a second embodiment of the vehicle position measuring device for electrified railway maintenance according to the present invention will be described. The vehicle position measuring device for electrified railway maintenance according to the present embodiment has substantially the same configuration as the vehicle position measuring device for electrified railway maintenance according to the first embodiment, but has the following difference: when there is a position of the last hanger, The above-mentioned hanger position is a reference and the processing area is limited, thereby speeding up the processing. Next, the configuration of the vehicle position measuring device for electrified railway maintenance according to the present embodiment will be described. Fig. 8 is a block diagram showing the configuration of the vehicle position measuring device for electrified railway maintenance according to the present embodiment. As shown in Fig. 8, the configuration of the vehicle position measuring device for electrified railway maintenance according to the present embodiment is substantially the same as the configuration of the vehicle position measuring device for electrified railway maintenance according to the second embodiment, but has the following differences: at the hanger position. In the case where the last hanger detection position is present in the detection processing unit 15, 155389.doc 201144108 transfers the last hanger position to perform the calculation of the above hanger field. The processing data of the auxiliary rail position measuring device for the electrified railway maintenance according to the embodiment will be described with respect to the tilting data of the hanger wire 103 and the wire 1〇1 and the position of the secondary processing zone in the processing area. The order of the electrified railway maintenance vehicle of the present embodiment is substantially the same as that of the first embodiment of the electric (four) slamming car center. However, the difference is that when the last position is added, the position of the hanger is added. The calculation process of the secondary processing area of the processing area is defined by the reference. Since the SIFT feature amount takes too much time in the calculation process, it is necessary to increase the speed of processing. The time required for the calculation of the SIFF feature amount is proportional to the size of the processing area. Therefore, if the processing area is available, the processing can be performed efficiently. Here, since the hanger 102 must exist between the hanger wire and the wire 1〇1, and there is a high possibility that the hanger wire 1() 3 and the wire 1G1 are inclined, the first hanger can be detected. 1〇2, you can limit the processing, domain. In this manner, since the secondary processing region can be limited in accordance with the inclination of the hanger wire 103 and the wire 1〇1, the processing speed can be increased. Fig. 9 is a flow chart showing the processing procedure of the position measuring device for the electrified railway maintenance of the present embodiment. As shown in Fig. 9, the processing procedure of the vehicle position measuring device for electrified railway maintenance according to the present embodiment is first in the steps. In 1(), the SI sensor is used to learn the template from the line sensor ^. In addition, the stencil image shown in Fig. 4 is an image in which a portion of one of the hangers 102 taken is cut. I55389.doc -13· 201144108 Next, in step P11t, the SIFT feature quantity is extracted from the template image to establish template data. 3. As shown by the black dot in Fig. 5, there are a plurality of SIFT feature quantities on the template image, and all the template data are created. Further, the method of obtaining the SIFT feature amount is described in the above Non-Patent Document 1 and thus the description is omitted here. Taking the test, in step P12, an input image is established from the line sensor i. In addition, as shown in FIG. 6, the vertical axis of the input image is height, and the horizontal axis is time. Next, in step P20, if there is a situation of the last hanger position, the calculation processing of the secondary processing area is executed in step P2I. In the case where there is no upper-personal frame position, the hanger position detecting process is performed in accordance with the entire input image in step pi3. Next, in step P21, the hanger ι 2 is detected from the j-th image, and the secondary processing region 5 is determined based on the hanger detection position 4 of the first image and the inclination of the hanger 1〇3 and the overhead line (8). . Specifically, as shown in FIG. 1A, when the inclination of the hanger wire Μ] and the wire ιοί are in the lower right direction, the hanger 1〇2 exists between the hanger wire 103 and the wire 1〇1, and thus exists in the hoisting It is more likely that the line ι〇3 and the line "I are inclined. Therefore, the secondary processing area 5 can be determined in the tilt direction of the hanging (four) H) 3 and the overhead line 1G1 based on the second image ... 4 . In addition, one of the dotted lines shown in A and B in FIG. 10 is used as a cut-off line of the circular image, and the left side of the one-point chain line A is used as the third image, and the point-chain line a and the γ-point chain line B are As the second image, the right side of the dotted line 8 is used as the flute image. ",,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, : 2 In step P15 'calculate the time from the detected hanger position and record the time β. Finally: In step Ρ16, if there is a situation of the last hanger position - 'hardening step Ρ 12~15 until the round When the image of the person does not exist, when the input image of the pair is finished, the processing of the vehicle position measuring device for the electrified iron maintenance of the present embodiment is performed in series. According to the present embodiment, the position of the vehicle for electrified railway maintenance is not only the performance of the electrified railway position measuring device of the third embodiment, but also the effect of the problem, and the treatment area is limited to the area 5 It is also possible to increase the speed of processing. [Embodiment 3] A third embodiment of the vehicle position measuring device for electrified railway maintenance according to the present invention will be described. In the electric railway transportation maintenance vehicle of the present embodiment, the electric railway 1/, the younger 2 M. , the maintenance vehicle_position measuring device has substantially the same configuration, but has the following differences _ ^ B + limits the processing area to In the case of the secondary processing area, the position of the hanger position is improved by expanding the secondary processing area 5. The virtual processing: The processing procedure of the vehicle position measuring device for electrified railway maintenance according to the present embodiment will be described. The 戥 dot chain line shown by A in Fig. 11 is used as the cutting line of the image, and the left side of the jersey is used as the second image of the second image. When the right side of the two-point chain line A shown in Fig. U is taken as the first image on the cutting line of the image, the SIFT feature amount will be different. Therefore, it may not be detected. The hanger 102 is taken out. Therefore, as shown in FIG. U, by expanding the range of the secondary processing region 5 to include the front image, that is, including a portion of the third image, the hanger can be detected on the cutting line of the image.丨〇 2. As described above, in addition to the effects of the vehicle position measuring device for electrified railway maintenance according to the second embodiment, the vehicle position measuring device for electrified railway maintenance according to the present embodiment expands the secondary processing region 5 as shown in FIG. As shown in the figure, even when the hanger 1〇2 is photographed on the cutting line of the image, the hanger 102 can be detected with high precision. [Industrial Applicability] The present invention relates to an electric railway maintenance vehicle position measuring device which is applied to the position of a side electric railway vehicle used for maintenance of a wire and a rail, etc. [Simplified description of the drawings] 1 shows the present invention FIG. 2 is a view showing a configuration of an electric railway measuring device according to an i-th embodiment of the present invention. FIG. 2 is a view showing a configuration of an electric railway measuring device according to an i-th embodiment of the present invention. A flow chart of the processing sequence of the electrified railway position determining device of the embodiment. Fig. 4 is a view showing an example of a stencil image of an electrified railway maintenance position measuring device according to a third embodiment of the present invention. Fig. 5 is a view showing an example of the SIFT feature amount of the stencil data of the electrified railway maintenance position measuring device according to the first embodiment of the present invention. Vehicle for vehicular vehicle 155389.doc • 16- 201144108 Fig. 6 is a view showing an example of an input image of the vehicle position measuring device for electrified railway maintenance according to the first embodiment of the present invention. Fig. 7 is a view showing an example of a corresponding point search processing region of the scary feature amount of the electric railway maintenance vehicle position measuring device according to the first embodiment of the present invention. Fig. 8 is a block diagram showing the configuration of a vehicle position measuring device for electrified railway maintenance according to a second embodiment of the present invention. Fig. 9 is a flow chart showing the processing procedure of the vehicle position measuring device for electrified railway maintenance according to the second embodiment of the present invention. Fig. 1 is a view showing an example of a secondary processing region of a vehicle position measuring device for electrified railway maintenance according to a second embodiment of the present invention. Fig. 11 is a view showing an example in which the secondary processing area of the vehicle position measuring device for electrified railway maintenance according to the third embodiment of the present invention is enlarged. [Description of main component symbols] 1 line sensor 2 illumination 3 processing personal computer 4 hanger position 5 times processing area 10 image creation unit 11 and 12 memory 13 template creation processing unit 14 SIFT feature quantity extraction processing unit 15 Rack position detection processing unit 155389.doc -17- 201144108 16 Time recording processing unit 100 Vehicle 101 Green 102 Hanger 103 Hanger wire 104 Structure 155389.doc -18-