TW201039617A - Automatic test method, test device, and test system - Google Patents

Abstract

An automatic test method, test device, and test system are disclosed. The implementation of the automatic test method includes utilizing a digital television terminal program to analysis a predetermined test sample and output image, so as to obtain the output image; comparing the output image with the predetermined test sample; if similarity between the output image and the predetermined test sample falls within a pre-set range, defining the result of test is passed; if not falling within the pre-set range, defining the test is failed. The method implement automatic test through comparing obtained output image with expected image, therefore manpower and undue testing time are both saved. Moreover, since pre-set range of similarity is the test standard, subjective factors of people are prevented in test and reliability of the test is enhanced.

Description

201039617 VI. Description of the Invention: [Technical Field] The present invention relates to the field of image processing technologies, and in particular, to a method, apparatus and system for automated testing. [Prior Art]

G Ο Digital TV software can be used to parse the information received by digital TV, and then output the received information. The output information can be displayed in the form of images through output devices such as TV. With the development of digital TV terminal software, the test workload of digital TV terminal software has multiplied. For example, a new TV terminal software branch is ready to be shipped out of the library or a TV terminal software branch is modified to prepare for delivery. The terminal software branch test needs to traverse all functional test cases, and the current test work is done manually. The test is sufficient and has a great impact on the quality of the terminal software. The process of 秘 秘 及 and the filling test can be decomposed into the following steps: (1) reading test changes to observe the contents of the test screen; (2) understanding the test object: according to the test knowledge of long time = accumulated test text Understand, know the test purpose of the test tooth =; (3) New test behavior: compare the expected test results in the test, and judge whether the test passes; in addition, if the record can be edited, mouse or face Interact the silk process or record test results. The skill person realizes the process towel of the present invention: the use of the human health test time-consuming test is a continuous 'manual test, the person needs to continuously pay attention to the test, a picture and the result' if the person's line of sight leaves the test screen, the test will There is an incomplete possibility, and the result of the manual test may be consequential, and thus the deviation may result in the uncertainty of the test result. [201039617] The technical problem to be solved by the embodiment of the present invention is to provide a method and device for automated testing. The system realizes the automatic test of the digital TV terminal software. In order to solve the above technical problem, the embodiment of the method for the automated testing provided by the present invention can be realized by the following technical solutions: · stealing ΐ 电 电 软 软 软 预 预 预 预 预 预 预 获取 获取 获取 获取 获取 获取 获取 获取 获取 获取 获取The acquisition_image_with the expected image of the predefined test instance is entered, and the range of the second r is optionally, the acquired output image is: the image; the acquisition and the real-world view of the domain of the virtual domain Contrast; the image of the domain is: in the predefined test instance, the image is marked by the area coordinate =··Wei TV terminal software solution _ _ try the instance and then input the instance, and continue to parse at least two different predefined tests: Comparing the input image with the __object axis image to the expected image of the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ image. 5 201039617 After analysis? Including the output image is compared with the expected image of the predefined test instance. If the comparison is displayed as the start of the recognition image, then the comparison is displayed as the end of the identification image, then the start of the comparison test; If yes: Analysis of the predefined test instance and round: After the image is also included 〇 〇 The acquisition of the output image also includes the output image in the display setting == display of the ruler interface image to determine the image. The acquiring the input image includes: acquiring the display position_image as an output image. Optionally, the image of the acquired display position comprises: finely outputting ^_ by means of optical word paste. Image = the expected contour of the wheel image obtained by II and the predefined test instance; the operation, the output image wheel is compared, and the expected image rim of the pre-suppression test instance is included to include the acquisition ___ The fineness of the fine range is set in the __________ 201039617 line comparison: the primitive value is represented by the color model; the included round image and the pre-image similarity are within the set range ^ _ wheel_image The side element of the hunger is in the set range Ο Ο The output image is the test result character image; the line to the wheel 1 _ 彳峨 side image into the ____ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The rounded image obtained by the rounded image and the job image similarity within the set range is a character indicating the passing. N-turn provides an automated test device, including: Example and 'used for the secrets of the electric woman's soft bribery _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The output image - *-, 1 trial instance J, = Bu Fan ===== phase as the result determination unit, optionally 'the acquisition round image is: Get the pre-sense test instance in the riding domain coordinates The circle of the marked area is shown in Figure 7 201039617 Image; 图像 The image of the 〇 field is: in the predefined test instance, the area above the area is marked by the area, and the number (4) depends on the number of instances of the final Wei body scale test. a different predefined test includes: comparing the rotated image with the expected image of the predefined test instance for comparison with the image of the package image ί. The acquisition touches the image corresponding to the image of the input image optionally The comparison unit includes: obtaining an output ^ image wheel 7 is operated by riding the image by the Sobel operator, and the expected surface of the circumference is defined by the test real package = the acquired output image is similar to the expected image in the set specification Fan =] 2 rounds out the similarity between the contour of the image and the contour of the expected image The comparison unit comprises: a second comparison subunit, a distribution scan, a statistical conformity, and an output image primitive obtained by the average color model representation; a color value of the primitive value; the primitive value is obtained by _ The similarity between the rounded image and the expected image is included in the set range 201039617: inner circumference: or, the expected color element value ratio of the rounded image is in the set range, the image is the fruit character map. The comparison unit includes: a character in the image and a predefined character subunit for acquiring the output image including the y J in the output map and the pre-fine image similarity in a set range Ο 的The characters that are rotated out of the image order are the characters that indicate the passage. The embodiment of the present invention further provides an automated test (four) system, including: a final image: FIG. 4; the output image is processed by a digital electrical test value; a system 4# example _; the image is sent to the test Really rely on ___ coffee definition line comparison, get the test result; if the obtained output image and the flip image her degree is in the set _, the thorn test ^ no, the test result is failed; the test result The server is sent to the database server for receiving and saving the test result; the front-end client terminal is used to obtain the test result saved by the database server. The mountain also 'the digital TV terminal software parses the predefined test instance and then outputs the image including: The digital TV terminal software continuously parses at least two different instances and continuously outputs the image; 〆 the output image and the predefined test instance The expected image is compared with the package 201039617: Line comparison fb The acquisition axis _ 轮崎 axis __ image into the optional 'This will be obtained from the output image image: μ off the face of the example Sobe _ operator to lose the butterfly image: the lion _ profile tender _ 财二 2; == Ο

The range of j includes the similarity between the acquired output image and the job image, and the similarity between the set image and the pre-fine image contour in the set-up image. The expected image of the test instance defines the image primitive for the average distribution scan, and the statistics conform to the pre- and the value ratio; the primitive value is represented by the color model; including the acquired wheel (four) image and pre-fine image She is in the set range; ^The ratio of the color of the image of the wheel _ image taken in the set range is the image of the result of the test; the image of the rounded image obtained by ί3 The expected image of the predefined test instance is compared with the predefined character of the obtained rounded image t; ~ The acquired image similarity with the look-ahead image is set in the scope of the field 10 201039617 Includes: The characters in the obtained output image are characters indicating the passage. The above technical solution has the following beneficial effects: automatic test is realized by comparing the collected output image with the expected image, which saves people's time and effort; in addition, the standard of the similarity range test is adopted, and the elimination is eliminated. The human understanding of the deviation, thus improving the test can be achieved [implementation] 实施 〇 拥 拥 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施 实施As shown in FIG. 1 , a method for automated testing provided by an embodiment of the present invention may include the following steps: Step 1G1 · Digital television terminal parses a predetermined _ test instance and outputs an image; Step 102 ··Get the above output image; 咖^(10). Compare the obtained output image with the line of the predefined test instance'·determine that the obtained output image and the expected image are ==No in the setting In the range, if yes, proceed to step 105; Step 104: The test result is passed; Step 105: The test result is failed. The execution subject of the step can be a job-feeding device. Of course, the test device can be connected in a communicable manner to achieve the above steps; the output image of the image and the expected image are obtained by the test result. The time and effort of automatically saving people; in addition, because of the adoption of a predetermined similar silk thread, the adoption of the test is fresh, eliminating the person _, thereby improving the reliability of the test. 201039617 执行 The execution body of step 101 and step 102 above may be a physical test device or a virtual test device, and does not affect the present invention. 4 *Step I. After 5, you can save the test results as 在 No, if you fail, you can save the detailed parameters of the output image. The above-mentioned acquisition output® is an output image of the region marked by the £ domain coordinate in the pre-defined example of the turtle; the image of the region is: the comparison range of the output image can be determined by the regional coordinates in the predefined 戦 instance. Zoom out to - a true two = thus reducing the complexity of image comparison 'At the same time can improve the image 3 output Lti bracket digital TV terminal software parsing predefined test instances and then the number of instances ^ * = = even transfer at least two __ Joy test: the ratio of the output image to the expected image of the predefined test instance.

Q Compare the acquired image images obtained above. The above embodiment of the image of the image corresponding to the image of the money, the above embodiment provides a continuous method of the above-mentioned plurality of predefined test real and the evening of a predefined test instance through a continuous test instance can be a 4, by different tests Purpose to determine; test results. - There are multiple instances of the test that need to be tested. Alternatively, the above acquisition takes turns to obtain the desired output image.匕 : 通过 通过 通过 通过 通过 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像 图像In the example, this will be explained in more detail. At least two different predefined test instances have an image; '..., w is used to mark the image, and one is displayed as the end of the logo. The image is compared with the expected image of the predefined test instance. The i image 'is determined - the beginning of the round test; if the 袖 采 袖 袖 , , , , 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 The example shows that the Lionheart passed the test. Bundle the mark ~ to determine the start or end of a round of testing. Included in the analysis of the predefined test examples and after the input _ image, the package will display the above-mentioned rounded image; the above-mentioned acquisition of the output image also includes: Display === (4) Fine and fine fiber image in the output The method of obtaining the rounded image includes: obtaining the image of the above display position as the output of the non-image output may appear as shown on the display of the position on the display device. The method is given to determine the output image through the ruler interface. For example, the method of the present invention has a method of obtaining the display position. The obtaining the image of the display position includes: acquiring the characters in the rounded image by means of optical sub-component identification. </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> Face IS (4) Feeding (4) Image and the pre-test of the default test instance, making a round-out _ line comparison, image contour and the expected image contour of the predefined test instance Ο The acquired output image The degree of similarity with the expected image is set in the set range is: the similarity of the contour of the out-of-round image of the rounded image contour including the difference is based on the difference, and the face of the color care wheel can be avoided. The influence of the slaves; in the comparative law. When the contour of the material is used, the image of the image based on the contour of the image can be used to calculate the average image of the output image elements obtained by the expected image. The ratio of the color element value of the color; the picture element 2^2 conforms to the pre-example test round in the set range [the test method of the image 3, which can more clearly care about the color of the output image] Similarity, for the ratio method. The image color-based test may be used to include the acquired image-to-pre-image similarity within a set range; the obtained output image has an expected color-element value ratio of 201039617 or the above-mentioned round-out chart. Like the test result character image; the input image obtained by iiif and the predefined real_predicted image will compare the characters in the obtained output image with the predefined characters. The character in the rotated image obtained by the output image and the expected image similarity in the set range is a character representing the passing. Ο = The software of the undefined software defines the difficulty of obtaining the test results. ^ : 11 : ^ τ ^ Trial 'for parsing the predefined test by the digital TV terminal software output image acquisition unit 202 for acquiring the above output image; trial-and-acquisition of the round-out image and the reservation _ The image is out of the second and the expected result is failed. . If it is passed, no, then the test surface image is obtained. Because the predetermined similarity range is adopted as the pass or not, the reliability of the test is improved by the understanding deviation of ‘:2: 卜, 15 201039617. Optionally, the obtaining the output image is: acquiring the wheel-out area of the area marked by the area coordinate in the predefined test instance. The second image is: the predefined test instance is marked by the area coordinate. 'You can narrow the comparison range of the rounded image to - a certain = thus reducing the complexity of the image comparison, but also can improve the image = output the digital TV terminal soft bribery scheduled _ try the example and then the example is even less The two __ financial test includes finding the output image and the pre-defined test instance _ _ Ο Ο = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = Test results for multiple predefined test instances. Optionally, as shown in Figure 3, the above comparison unit includes: Counting 'round: testing the output image through the Sobel operator ί Example of output image contour and predefined 16 201039617 Intrabacteria include = description of the acquisition_image_image and _image similarity within the range of the set output image contour and the expected setting; the similarity of the image rotation is within the difference of Ο = = 3⁄4 like the wheel r: mode, the color r_m can be avoided: or the above comparison unit includes: · flat: the similarity between the output image obtained in the sub-area and the expected image is in the set range; The expected color element value ratio of the obtained output image can be more clearly related to the similarity of the output image color in the test f formula of the set standard test color, for the ratio mode. When α is used, the image color based test or 'the above output image is used as the test result character image; the above comparison unit includes: pre-"; the character and the surrounding character in the output image will be acquired. The characters in the rounded image acquired by the set image in the image of the paste and the paste image are the characters that pass through. 17 201039617 In the analytic theory, the test result of the digital TV terminal software example 'the result of the above test = the test image of the meaning table &amp; threat risk = the character to directly obtain the output image and the pre-form As for the comparison, you can get the _! test wire, so that the 戦 细 获 获 就 就 就 就 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 本 采集 采集 采集 采集 采集 采集 采集 采集 采集 采集 采集 采集=:^: rotation fruit = library feeding machine; n receiving results sent to the data library 4G3, used to receive and save the above test results · results front-end user terminal, the remaining test for the accumulation of the deposits Standard, eliminates the 'output vertical 赖 thief _ _ example and then instance number fiber (four) 'surgery _ definition test 18 201039617 The above will take out the image with the pre-defined test carefully include: the expected image of the gas case comparison 2 get The expected image corresponding to the output image of the above-mentioned rounded image. The above embodiment provides a continuous measurement method. The above plurality of predefined test instances can predefine the test instance by successive test instances. - test = ^ _ test purpose to determine; test results. , the period diagram of the various instances that need to be tested ίίί and the predefined test through the Sobel operator pair to rotate the image 靡; the above-mentioned rim instance of the pre-line comparison; including the ^ _ _ _ _ _ 嶋 settings Fan's input is expected to be within the range of image wheel circumference setting; #similarity is in effect; in comparison of differences, test mode 'can avoid color to lose == yuan format, the formula caused by test results.俺 __ tree _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The similarity in the set range 201039617 includes: The rounded image obtained within the J-shaped π value ratio can be tested more clearly in the set range. The test method of the output color is more... The similarity of the color of the output image of the ,, for the ratio. In the case of Bashan, the image color-based test can be used as the test result character image; Ο 〇 the obtained image is compared with the expected image of the predefined test instance (4); The character in the output image obtained by the rounded image and the expected image similarity in the set range is a character representing the passing. After the solution to the script-based test method 'digital TV terminal software', you can give a test method for parsing the above-mentioned predefined 3 丨 script. When the image is compared, you can compare the if image with the predefined result image. τ is less difficult to obtain the results of the results of the Lai (four) test results. Fourth Reality, (4) Implementation A more detailed embodiment of the above-described first embodiment and the second embodiment is given. In this embodiment, the example (CASE) is a test case (also referred to as a test instance), and the test case in the embodiment of the present invention can be expressed in the form of a CASE picture, so the CASE picture mentioned in the text can be considered as a CASE. The device for parsing the test case can be a hardware digital television terminal, for example, a set top 20 201039617 box, or an analog digital television terminal, such as an emulator. The device to be tested: it can be a digital TV terminal software, and the digital TV terminal software can be embedded in an operating environment. The above environment can be a hardware digital television terminal or an analog digital television terminal, so the device under test can be understood as : The physical device can also be an analog device. Steps 101 and 102 of the first embodiment of the present invention can be understood as accurate identification of test cases, including two cases; f (1) Static CASE number identification (can be understood as identification of a single test case): ° 〇 s stomach The static CASE number identification is a method for extracting the CASE number for a fixed CASE picture. Each valid CASE picture must have a number. The identification is implemented as follows: Step 1 · Read the coordinate position of the CASE picture number area, the above number The area may be a fixed position input by the user; Step 2: intercepting the CASE picture number area sub-picture according to the coordinate position; Step 3: performing character recognition on the numbered sub-picture by using 〇ptical Character Recognition (OCR) technology Ο Step 4: Analyze and verify the obtained character recognition result to determine whether the coding rule is met. Through the above steps, the accurate recognition of a static CASE is completed. (2) Dynamically and continuously issuing CASE (can be understood as the identification of multiple test cases): In the test environment, the parent CASE can be automatically sent, received and tested through the terminal software continuously and dynamically at fixed time intervals; Continuously issued CASE screening can be _ for riding CASE _ up to accurate downsampling' and then the dynamic process of identifying the "Static (10) numbering Lang" method of 21 201039617 T +, the implementation can be: Step 1. Every The received (10) picture is sampled once every 2 seconds. The period of the above sampling is set to other _, for example, seconds. The invention is not limited, and 疋J 乂 compares this (10) picture with this image processing technology. The image of the upper-person sampling 1 is compared with the image, and only the front and back of the two CASE images are compared. The result is different (if this is the first sampling) The last sampled picture is definitely different from the result. Otherwise, the CASE number is not the picture arrival sign, and the current time is the time when the new case picture arrives. If the comparison result is the same, Whether the difference between the current time and the last new CASE picture to = time is greater than the number is unchanged _, the "CASE compilation presents a constant time" is calculated by the "CASE interval jump time, calculated. If exceeded, the current can be determined (10) The picture is a valid (10) picture, then = it is identified by the static CASE number identification method, and it enters the test process &gt; 711 • otherwise; otherwise, the above steps 1 and 2 are repeatedly executed. Example: Assume that the CASE page jump time (1) is 8 seconds, and a solid CASE page jumps from the beginning to the whole sister to the end of the page. It may take 1~6 seconds (8). When the new CASE page is detected, it can't immediately capture the picture as a CASE picture. At this time, the CASE picture is very = the month b is not integer. Since the detection sample is performed once every 2 seconds (evaluation), from the time when the new CASE page is detected to the time when the page is finally fully presented (τ) &amp; CASE The number is always the same (can be understood as: from the detection of the new CASE, it is necessary to wait for a period of time τ, this time is called CASE number = change time, you can capture the map as a valid CASE Picture), the experience summary is roughly: τ = (I - w) χ θ.6; 22 201039617 The method of comparing the test comparison of the steps 1〇3 in the first embodiment of the present invention can be divided into three types: the use case based Color test and script-based test, their test step 1: read the label area of the CASE test attribute; the same as step 2 · traverse the label area of the CASE test attribute: If the label area is required to be based on the contour test : Then 'first intercept the (10): after testing the marked area of the picture, read the standard of the CASE corresponding number, intercept (10)

The label area sub-picture; finally 'use the image processing technology to compare the label area sub-picture of the ^ test picture with the label sub-picture of the CASE standard picture, and if the comparison result exceeds the set expectation pass threshold, judge the ° ° to test Pass, otherwise not pass; if the marked area is required to be based on color testing:

First, the labeled area sub-picture of the CASE test picture is intercepted; then, each picture element of the labeled area sub-picture of the CASE test picture is scanned to determine whether the color value of each picture element meets the expected value; and finally, the statistics meet the expected color value. The proportion of the element point 'if the ratio result exceeds the set expectation threshold', the CASE test passes, otherwise it does not pass; if the label area is required to be based on the script test:, Bay 1 first intercepts the fixed CASE test picture The script result displays the area sub-picture; then, the result string of the sub-picture is identified by optical character recognition (OCR) technology, and if the result string is "PASS", it is determined that the CASE test passes 'otherwise not passed; step 3 : Save test results. More specifically, the implementation process of the above embodiment may be divided into six parts: (1) making a test case; 23 201039617 Each CASE of the automatic test may be rendered by one or more web pages: during the automatic test process Each CASE is treated as a picture, and the test is transformed from the viewing terminal software parsing presentation mode to the image processing mode of the rendered picture picture, so that each picture is an independent CASE. Here, this CASE picture is called a standard picture. For the convenience of uniquely distinguishing each CASE, the Hypertext Markup Language (HTML) is embedded in the lower left of each CASE. The digit code is identified, for example "010101";

G 〇 ▲ automatic test CASE production has two main aspects: one aspect is the automatic f CASE standard picture selection, after being used as a test case standard picture to rely on human experience to verify and confirm the correct rendering, stored as test ίΐί image. Take the numbered "Proposal 101,, ❺ (10) as an example, the test case test = "In the source code towel inserted, the view ^ shirt visible", expected = if the following only - line code instead of two lines, left If you try to pass the "visual observation of the naked eye, it does match the expected phenomenon", you can store the picture as a fan-free image in the 11 places for automatic job comparison. The other side is referred to the following section (2): (2) Labeling test cases; this test case can have specific test attributes and test purposes, which can be expressed in text or pattern. m for people ~ job, _ person's direct _, can be the test attributes and test purposes, and get the test results. But what about reading the lion's job? Computer measurement and test purposes include::=== 201039617 From the fine system, here you can satisfactorily (10) picture coordinates parameterized storage in the library for automatic shame comparison; 'master (3) test cases issued; After the two steps of the test are ready for the test case, you can start to enter ❹ 〇 iiH 'as mentioned above, each case picture has a corresponding broadcast two i week page) 'To ensure that the test is continuous, you need to All the case pages are stored in the 'then the terminal software through the hypertext transfer protocol (Hypertext Transfer Protocol? HTTP) / mm ^ (Internet access to these prepared CASE pages, in order or randomly parsing the broadcast, the mother - a CASE The same time length of the webpage is the same as the time of the webpage. These can be changed by the navigation setting. It is divided into the side-by-side trials. First, enter the jump setting page. Before the next sCASE page, you can pre-connect the parent CASE page. The previous and next jump links, as long as you enter any CASE page, the automatic jump is triggered by the embedded CASE page. The automatic jump event triggers the next jump from the CASE page. It can be flipped by the remote control of the work industry; in addition to the way of sending CASE over the HTTP/IP network, it can also be passed through the object carousel generator (〇bj.ec1: Car〇usel

Generator, 0CG) treats the CASE page as a complete object carousel (Object Carousel '0C) application broadcasts. 'Digital TV terminals receive cable TV in the Hybrid Fiber-Coax (HFC) network (Community Antenna Television, The CATV signal can also achieve the same purpose. In order to accurately detect the start of the test, before the test case is sent, after setting the delivery page, a flag picture indicating the start of the test can be issued. When the test case is received, it can be detected whether the test actually starts: the detection start is marked by The OCR technique reads the image setting area, and when the recognition result is "START", it is considered that the start label has been detected. Measure 25 201039617 The test case is issued after the 'Phase 264 encoder is received, it may become the following shape. On the left and right, there are extra black borders in the three directions. The reason is that the round-out resolution of ^ (704x576) and The round-trip parsing (_^&lt;526) of the digital TV terminal software is different, resulting in non-coincident areas. For different ^'s, the size of the non-coincident border is not certain, which will cause the front production (^ASE) There is an error in the positioning of the marked area of interest, because the actual 椤, and the main area parameter reference values are marked on the premise that there is no border effect, and the error is caused by the uncertainty of the frame. The system is processed by the calculation method of the guide ruler. The specific implementation is as follows: Ο Although there are black borders on the top, left and right, the position of the marked area can be accurately located as long as the exact position of the upper and left black borders is obtained. There is a clear dividing line between the = black border and the terminal outputting the effective image. Finding this dividing line will accurately know the border size. We developed A specific ruler detection algorithm is used to test the position of these black borders, thereby eliminating the machine image size and the capture card input image size. The pairing is automatically created = 0 (4) test case reception; CASE page passes through the digital TV terminal software. After parsing, the output is a digital video recording of the system. The trick is to input the video directly into the TV' and then observe it through the human eye. &amp; The automatic test system is based on image processing H. The touch signal is processed into an image for processing. Here, the 264 encoder is used to directly receive the output of the digital TV terminal and then output the sampling key + map for the 264 encoder. The video frame rate output by the H·264 encoder. Can be throwing frames / second, which means that 25 images per second are continuously output, automatically test the secret &lt; not to take all the pictures for processing, because consecutive multiple _ like may be the same picture content '- a CASE After the page parsing is sent out, it is necessary to automatically jump to the next page after the set time. In the time period of the money, the encoder 26 201039617, each of the 1 is almost the same picture. Rong, such a valley, from which to choose the complete (CASE page all elements are taken into the screen = ί ^ ^ ^ _ _ like to be treated to reach the best two = complete, representative - _ like the key ΐ ΐΪΐ: i;: _, here we use the technology of image lens segmentation: two adjacent (four) jump interval is 8 seconds, after the last CASE analysis, the final output of the 264264 encoder is ··· 〇Ο 〇=2^ After the Ϊ 264 264 264 encoder finally turns out as: 庠 column / 'Caf = the actual two lion images after the Η · coffee code output of the order 歹 i疋oloioi is not visible η# And then display "〇1〇1〇2"; assume that frame 0 is the ith frame of CASE "010101", the nth frame is 〇! The last frame of 0101' is the CASE 201 at the beginning of the η + 1 frame. After the n-frame is passed, the next case image will appear. Suppose the output frame rate of the 264 encoder is 25 frames/second, and the CASE page is sent at 8 seconds. Then η should be equal to 25 X 8 This digit is not an exact digit. During the actual delivery process, the time interval is not necessarily strictly accurate, and the terminal is soft. Resolution is not necessarily accurate precise time every time, so I can not exactly let the computer know that a picture from the next frame image has changed 'but due to the continuous multi-Cloth! The jj贞 image may be a picture content, so that the computer can accurately know that the image of the previous second and the previous second has changed, or the image of the first two seconds and the last two seconds has changed. The image is taken, and then the image comparison method is compared to detect whether the image has changed. The specific sampling detection process is as follows: sampling every interval of 2 seconds, and then comparing the image of the current sampling with the image of the last sampling. The similarity, if the similarity is lower than a certain threshold, it is considered that the image of this sampling is different from the content of the last sampled image, that is, a new CASE image arrives, but the new CASE picture at this time is not yet It must be complete, because a CASE page is parsed by the terminal software until the rendering complete has a time delay of 27 201039617, so after detecting the arrival of a new CASE picture, it still needs to wait for a certain delay time to ensure that the tested CASE picture is The complete, specific delay time is determined as follows: (CASE page delivery time interval (default seconds) -2) X 〇. 6 ; 〇〇 Compare two images is No similarity can be used to compare the gray image of the two images Sobel (S0BEL) operation sub-image is similar, take the above two CASE pictures as an example 'calculate the S0BEL operator image of its gray image; then one by one From left to right, the image is compared with the two s〇BEL operator images from top to bottom. The similarity is 61.62%, and 80. 00% is set as the basic similarity threshold of the two images, so this is detected. The content of the two images is different, in order to improve the detection efficiency and the accuracy of the detection; the degree 'can narrow the comparison area to the numbered area, since each valid CASE will embed a six-digit code, six adjacent CASE The bit coding is definitely different, and the location of the entry is fixed; the binary phase ^ ^ two CASE six-bit encoding is definitely different, but generally only one number is different, in order to reduce the proportion of the same part too much The detection accuracy is not high, you can = step into the 7&quot; step to narrow down the comparison area, and directly take the coded post-three-dimensional area for detection. The above-mentioned self-small test system can accurately detect the received integers. ,Have Table of CASE. (5) test case processing; test case processing is also graphical month processing, the process of test processing can be roughly divided into four processes: = 戦 (10) picture: how to encode the image six-bit field bit ^ ^ silk Learning the word lang (10)) The L on the image of the techno image is changed to the 疋 疋 ' Whether it is the data storage or the program identification, the six-digit can make the automatic test system know that the positive picture is produced. · According to (10) 'Number in Fixed storage of the standard location of the SE standard picture to read the standard of the CASE Figure 28 201039617 The purpose of the film is to compare the test picture with the standard picture; the data focus on the test area: according to the _ number in (10) test processing with test label area parameters The purpose of the information is to allow the test to be ==, GASE, and the financial job area to deal with: the same - a CASE test the second main area for testing, each of the marked areas can be different concerns 1 'we each individually tested; The test processing is divided into three types of tests: A. Based on the graphical round-robin test, · Ο Ο 于 于 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 要求 ' ' ' ' ' Sex is the part of the box. When the box is round and the five-pointed star, the test passes. For the automatic test: =: What is the detection of what is called - a dot and a five-pointed star, how big a dot and a large star are met. How do you position the dots and the five-pointed star? This requires the use of the standard picture in the previous production of CASE: The quasi-picture can be selected in advance by the person I, as long as the actual test, ▲ received the picture is exactly the same as the standard picture, or very similar, even if passed. Here, using the edge detection technology, the round information of the test picture is obtained from the rim information and the standard picture. The first set of this picture is the attribute attribute_Note area gray image S0BEL operation sub-process processing real test attribute round gallery information . Then compare the information of the fresh job attribute light and the actual test attribute, the comparison method can be from left to right, from top to bottom, and the similar proportions are obtained from the top to the bottom, for example, the similar proportion is 86·15%. Set the pass threshold to 8〇/6 ' so this use case test passes, otherwise it does not pass. The advantages of contour-based testing are: avoiding the effects of color differences; avoiding the effects of light intensity; avoiding the effects of primitive formats. B: Based on image color test; Sometimes the attribute required to test is color-sensitive' (eg: CASE 29 201039617, the test attribute is the two red line inside the box, if the color of I and point is red ' = job _ For the self-respecting of the attributes of the marriage of the color of the color with the light intensity, terminal equipment display, Η · 264 code collection and other reasons of the same degree of change. Take the above CASE as an example. Its (10) is issued in the source area, in the test attribute! And the point color is described by the three primary colors (10) as (255 'G, G) ' while the actual receiving point and the I color value may become: (48, 49, 31) and (10), 4, 22), and the yan ί ΐΓ Change, and - the name of the color is human natural language 1 == _ meaning, the same name defines the color in the visual effect

G 〇nil ), (255,132,132), (174)〇^)^255^0, ,64,124) are all considered red, and the RGB values are different, so that they cannot be compared _ is the same, Or compare _ can not be compared between the difference of the multi-shot color can be compared between the 'automatic test system will be on the production", break from the +. RG color value space to the color model [Hue (color ίH (saturation Degree), Value (purity), _ space, in view of the change in the environment according to the specific circumstances, the automatic measurement of the quilt is less woven and several colors of the needle ====; =? divided into human natural test system The color test surface is compared with the symbolized color value and the gradation color value. The color recognition problem can be recognized. (4) The value is matched, and the reading can solve the area label of the area of the color. The contour is not based on the contour test attribute to mark the entire labeled area. As a whole, whoever "tests the color of the area of the cover may be close to other color blocks. For this, Yan 30 201039617 asks to mark the area of the full side of the area is the expected color value, but can be in the marked area Average sampling (10)_yuan points, weaving the color value of the 100 side points 'most Counting the 100 primitive points to determine how many primitive points meet the expected color value, and set a pass threshold of 5, that is, when the marked area is recorded, the body is equal to 5% of the Lai Yuan points. If the test attribute is passed, the same can be applied to the color exclusion test, that is, the "inverse color" test. As long as the desired color does not appear in the marked area, the test attribute is tested by 'limitation is the opposite, when When there are 5% or more of the pixel points meeting the expected color value, the test attribute test fails; C: based on the script test;

D Sometimes CASE can detect the success of page parsing directly by using HTML script language. It can be combined with HTML script and image processing to achieve the test purpose. For example, the test attribute of CASE is the evaluation function of test text box, as long as the title is detected ( Title) The defined value is "test pass". This test passes, and the detection of the title definition value can be completely executed by the HTML script. In order for the automatic test system to quickly recognize whether the HTML script function passes, it can be executed by the script after the page. The fixed position is embedded with a fixed recognition character, which can satisfy the consistent character recognition mode of the automatic test system. Here, the "PASS" is output at a fixed position on the right side of the numbered area, and if the script performs ^3 work, the output "failure" (FAIL)", the specific script page execution example is as follows: &lt;P&gt; Test attribute: title attribute in br tag&lt;/p&gt;&lt;P&gt; Test function: set or get object consultation information &lt;/ p&gt;&lt;P&gt; Tested phenomenon: title is just a description here, does not implement any function, To get its defined value &quot;test passed" then _ try by &lt;/p&gt; ' 31 201039617 &lt;br id=”br_id&quot;title=&quot;testby&quot;> title = &lt;input type=&quot;textl&quot ; id=&quot;textl&quot;Xbr&gt;&lt;br&gt; • · _ 〈SCRIPT LANGUAGE:&quot;JavaScript&quot;〉 var testvalue = document. getElementByldC' textl&quot;). value; if(testvalue=='test passed') o {

Document· getElementByldC cresult’)· innerHTML=,P ASS,; } else {

Document. getElementById(, cresulf ). innerHTML=, F AIL, } 〈/SCRIPT〉 羼• · · ' The automatic test system identifies the area of the predetermined test result by the MR as "pass, even if it passes, otherwise the output is Mil 'or no output is considered to pass. (6) Save test results; 32 201039617 If the test passes, you can save only the test case number, test type, test time 'If the test is not completed, you can also save the marked area that does not pass. The result value of the body and the CASE image at the time of the test failure can be easily checked by the tester afterwards. ',讦,,,. The whole automated test system can include six system components. Figure: α video capture server 501, analog capture server 502, test processing server 503, database server 504, central platform server 5〇5, foreground configuration terminal 506; its application, video capture server 501 is responsible for Digital TV terminal signal acquisition, and after the lens split processing, the new valid CASE picture is passed to the test processing 503; ^ analog acquisition server 502 is responsible for the digital TV analog terminal output acquisition, and after the lens segmentation process, the new valid case picture is transmitted to the test processing server 503 through the network; video capture server 50 simulation The acquisition server 5〇2 performs the above-mentioned automatic test process (4) “receiving the real-life example”; it can be understood that the video capture server 501 and the analog acquisition server 5〇2 may both exist, or only one type exists; The processing server 503 receives the CASE from the video capture server 5〇1 or the analog server 502 and tests the CASE, and saves the final test result on the database server 5〇4 through the network. The processing server 503 performs the above-described automatic test process (5) "test case processing,". The central platform server 505 is responsible for managing the correspondence between the plurality of video capture servers 5〇1 and the analog acquisition server 502 and the plurality of test processing servers 5〇3. It ensures that any one of the acquisition servers can be effectively added to the test system. Obtaining the processing of the feeding device resources also ensures that any one of the test processing feeders can be added to the test 33 201039617 system can effectively allocate resources to each video capture server 5 {Π, analog acquisition device 5〇2, of course, the central platform server 5 〇5 can be used to ensure that all deployment services that participate in the test system can communicate effectively; = station configuration terminal 5G6 is responsible for the entire automatic test secret test case buried, Beixun management, test management and test results viewing management; Ο 〇 次^Status> Management includes basic information management, such as: test case test equipment information (ie, classification description information of the tested terminal equipment), if case information (ie test case database information), test channel information (ie dimension) , 5,, information) and user information, feeding device management information, stop test iiU including various operations of the test 'such as: start test Test, look, test results, etc.; front-end configuration terminal test ^ special i ^ test system information configuration, test various operation entries and β, σ fruit look at the window, do not participate in the test process. Fengkou f also includes: Internet 507, TV network _, network port and serial port conversion Is 509, router 510, H. 264 capture card 511; = medium network 507 for video capture feeding device measurement, simulation Collecting the ship's feeding service test, ^ί test office, 5〇3, database ship 11 504, medium: platform s, data parent channel, video capture server 501 picking shell through H.264 capture card 511 After the analysis, the routing_SI card and the analog mining can provide the recommended acquisition 34 201039617. The description of the above embodiment is only used to help the I and 3 methods and their core ideas. At the same time, for the general technology in the field. The limitation of this hairpin thought, in the specific real money style and the range of care ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄ Method flow intent, FIG. 2 is a schematic structural view of a device according to a second embodiment of the present invention; FIG. 3 is a schematic structural view of a second embodiment of the present invention; FIG. 4 is a schematic diagram of a third embodiment of the present invention; Figure 5 is a schematic view of the fourth embodiment of the present invention.

[Explanation of main component symbols] Parsing unit....................201 Output image acquisition unit.........202 Contrast unit....... ..........203 Result determination unit.............2〇4 arithmetic unit.................301 First Contrast Subunit...........302 Second Contrast Subunit........303 Third Contrast Subunit........... 304 Acquisition Server...............401 Test Processing Server...........402 Database Servo jj艮........ .....403 35 404 201039617 Front-end client video capture server...........501 analog acquisition server...........502 test processing server.. .........503 Database Server.............504 Center Platform Server...........505 Front End Configuration Terminal... ..........506 Internet.................507 〇TV Network.............. ...508 network port and serial port converter.........509 router...................510 H. 264 capture card.... ..........511 〇36

Claims (1)

201039617 VII. Scope of application for patents: A method of automated testing, including: The finalization of the escaping of the pre-defined 峨 实 顺 输出 输出 ' ' 获取 获取 获取 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' When the image is entered, the similarity between the image and the pre-image is set. If the result is 'No', the test result is not passed. 2. The method of claim 1, wherein the acquiring the rounded image is: the image; obtaining the rounded image of the area marked by the area coordinate in the predefined test instance is: the area coordinate field in the predefined test instance In the expected image display area test, the digital TV terminal soft-sense parsing is predefined ❹ The example is connected with the pre-defined test of the 杂 杂 杂 、, and the output image is compared with the expected image of the predefined test instance. The method of obtaining the output image includes: ▲ the output image required for obtaining the image of the image. Method, _ wherein the at least two non-predefined tests are displayed as the start identification image, and one parsed is displayed as 37 201039617 end identification image; wherein the comparison of the rims includes: · _ like _ definition test If the contrast is shown as the start of the logo image, the expected image of the example determines the start of a round of trials if the needle is sharper than the stem. The right contrast is not the end marker image. 6. As described in claim 1. Method, which ends. 〇 Get the display _ as a ride out image. 7. The method of claim 6, wherein the obtaining the image of the display position comprises: ▲ acquiring the character in the input image by means of an optical character. 8. The operation of the rounded image is performed by the Sobel operator. The ratio of the output image contour to the pre-fine image of the pre-coded image is like the image of the indefinite image. Comparing the similarity of the wheel_image corridor with the expected pattern or comparing the acquired output image with the expected image of the predefined test instance includes: performing an average distribution scan on the acquired rounded image primitives.符38 201039617 Combined with predefined definitions; 4 疋 疋 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过 通过Comparing with the predetermined output; the obtained word j is a character in the rounded image obtained by the passed word ^. An automated test device comprising: a virtual digital television terminal software parsing - predetermined _ 〇 The output image obtaining unit is configured to acquire the pre-:= line of the wheel_image; and the image of the area indicated by the area coordinate in the pre-defined test instance of the acquired round-out image and the predefined test instance • as clear The device of item 9, wherein the digital television terminal software parses the predetermined 羲39 201039617 test instance and then rotates the image comprising: the digital television terminal software continuously parses at least the instance and continuously rotates the image; two different predetermined 羲Trial package: the output image is compared with the expected image of the predefined test instance, and the g-acquisition image is outputted with the expected image corresponding to the output image. The comparison unit includes: obtaining Si; wheel: performing an operation on the wheeled image by the Sobel operator, and the example of the output® image wheel axle is predefined, and the output image and the expected image obtained by J are included in the trial. The image similarity is set in the set of Fan Park ^ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ In the second example of the figure, the TG value of the figure is represented by a color model; the expected color element of the rounded image of the rounded image with the expected image similarity within the set range The value ratio is set to 201039617 or the output image test result character image; the comparison unit includes: a pre-ratio for taking out the rounded image obtained from the character f in the output image and the expected image The range includes: Similarity is set in Ο η - Record "The character in the acquired output image is the character that represents the pass. 13. - A system for automated testing, including: Acquisition of the Wei device for obtaining the output map For example, the visual terminal software parsing pre-defined test instance is obtained by the over-digit electrical test processing server; the output image is sent to the test station _ service H' for comparing the expected images of 2 cases, and the test result is obtained. Out of the image and the flip image, the degree is within the set axis, ^ test ^ no 'the test result is not passed; the test result is sent to the library = fruit and is used to receive the database server to send The test result is used by the front-end user terminal to obtain the test result of the data pasting device. The system of claim 13, wherein the digital television final test instance then outputs an image including: · miscellaneous analysis Continuously parsing at least two unconstrained test instances and continuously outputting images; 201039617 Included • This will convert the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The system of claim 13, wherein the expected image of the test instance is compared, wherein the obtained output image is compared with the predefined output image by the Sobel operator. The input wheel is predetermined axis =: 3⁄4 circumference = the acquired wheel image is similar to the expected image similarity in the set fan setting image contour similarity in the line of the rounded image and The expected image of the predefined test instance is obtained by the rounded image obtained by the predefined_color_yuan value^1仃average* cloth scan, the statistical character, wherein the instance value is represented by the color model; The image of the output circle is obtained from the expected image similarity with the expected image. The ratio of the expected color element value of the output image is set in the image result character image; the output will be obtained. _ pre-image _ includes: ratio; in the image Predefined characters for character pairs, three image and the expected image similarity within the set range 42201039617 comprising: obtaining the output image is a character by character. 43