JP6432296B2 - Device and method for detecting zebra crossings in images - Google Patents

Description

  The present invention relates to the field of image processing, and more particularly to an apparatus and method for detecting zebra crossings in an image.

  In recent years, driving assistance systems have been rapidly developed. The driving support system can identify the environment before an accident occurs, and is therefore an effective tool for improving traffic safety. There are many sensors in the driving support system, which can detect information for preventing accidents. Currently, by using cameras and a wide variety of image processing algorithms, the driving support system can identify features in more traffic scenes and enhance its intelligent analysis capabilities.

  Embodiments of the present invention provide a zebra crossing detection apparatus and method. By calculating the visual features of the image based on the edge features of the detected image and determining whether zebra crossings are present in the decision image based on statistical results for the visual features, the vehicle and zebra Realizes zebra crossing identification when the distance to the crossing is far (large), that is, can predict the advance of zebra crossing ahead, so traffic safety and real-time detection of zebra crossing Accuracy can be improved.

According to one aspect of an embodiment of the present invention, a device for detecting zebra crossings in an image is provided, wherein the device comprises:
A first calculation unit for calculating visual features of the image based on detected edge features of the image, wherein the visual features include a saturation feature, a spacing feature, and a luminance difference feature And a first processing unit for determining whether the zebra crossing is present in the image based on a statistical result for the visual feature calculated by the first calculation unit.

According to another aspect of an embodiment of the present invention, a method for detecting zebra crossings in an image is provided, wherein the method comprises:
Calculating visual features of the image based on the detected edge features of the image, the visual features including a saturation feature, a spacing feature, and a luminance difference feature; and based on a statistical result for the visual feature Determining whether zebra crossings are present in the image.

  The beneficial effect of the present invention is that the method and apparatus in the embodiments of the present invention achieve effective identification of zebra crossing when the distance between the vehicle and the zebra crossing is long, and the speed and effectiveness of immediate detection of zebra crossing. And avoiding false detection of zebra crossings.

2 is a flowchart of a method for detecting zebra crossing in an image in Embodiment 1 of the present invention. It is a flowchart of an example of step 101 in Example 1 of this invention. 6 is a flowchart of a method for detecting zebra crossing in an image according to Embodiment 2 of the present invention. It is a figure which shows the region of interest acquired at step 303 in Example 2 of this invention. It is a figure which shows the edge feature point in the edge feature image in step 306 in Example 2 of this invention. It is a figure which shows the sampling point in step 307 in Example 2 of this invention. FIG. 6 is a configuration diagram of a device for detecting zebra crossing in an image according to Embodiment 3 of the present invention. FIG. 6 is a configuration diagram of a first calculation unit 701 in Embodiment 3 of the present invention. FIG. 6 is a configuration diagram of a first processing unit 702 in Embodiment 3 of the present invention. FIG. 6 is a configuration diagram of a device for detecting zebra crossing in an image according to Embodiment 3 of the present invention.

  Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the accompanying drawings.

  Embodiment 1 of the present invention provides a method for detecting zebra crossings in an image. FIG. 1 is a flowchart of a method for detecting zebra crossing in an image in Embodiment 1 of the present invention. As shown in FIG. 1, the method includes the following steps.

Step 101: Calculate visual features of the image based on the detected edge features of the image, the visual features including a saturation feature, a spacing feature, and a luminance difference feature;
Step 102: Determine whether zebra crossings are present in the image based on the statistical results for the visual features.

  In this embodiment, when zebra crossing is detected, one image may be obtained first. For example, one image of a road ahead may be obtained using an in-vehicle camera. Thereafter, image processing is performed on the image to detect whether zebra crossing exists.

  In this embodiment, an edge feature of the image can be detected, and then a visual feature is calculated based on the detected edge feature, and the visual feature includes a saturation feature, an interval feature, and a luminance difference feature. Thereby, compared with a prior art, zebra crossing can be detected more correctly.

  In this embodiment, in order to improve detection efficiency, the method further includes the following steps.

  Step 100: Prefix processing is performed on the image. Step 100 is an option.

  Among them, first, the detection waiting image can be normalized to a standard size, noise removal processing is performed on the normalized image (for example, by using a Gaussian smoothing filter), and after the noise removal The image may be converted into a gray scale image, or a region of interest (REGION OF INTEREST, ROI) in the image may be detected as a detection waiting image. Since the pre-processing as described above can be realized by adopting the conventional technique, detailed description is omitted here.

  FIG. 2 is a flowchart for obtaining visual features according to the first embodiment of the present invention, and shows an example of step 101. As shown in FIG. 2, the method includes the following steps.

Step 201: Select one line of data in the image;
Step 202: Detect a plurality of edge feature points in the selected row of data;
Step 203: The center of two adjacent edge feature points among the plurality of edge feature points is set as one sampling point in a selected row, thereby obtaining one or more sampling points;
Step 204: Calculate a line of visual features in the selected image based on the plurality of edge feature points and the one or more sampling points.

  In step 201 of this example, any one row of data in the image may be selected, and the present embodiment is not limited to this. In step 202, an edge feature of a row of data in the selected image can be detected by a Canny operator edge detector, a Sobel operator edge detector, or a Roberts operator edge detector. The above is an exemplary description, and the present embodiment is not limited to this. In addition, the execution order of step 201 and step 202 is not limited to before and after, for example, the edge feature in the image may be first extracted, and then one line of data in the image may be selected.

  In step 203 of this example, the center of any two adjacent edge feature points of the plurality of edge feature points can be set as one sampling point in a selected row, whereby one or You can get multiple sampling points. Hereinafter, the case where the sampling point is the center of two adjacent edge feature points will be described as an example. Note that the sampling point acquisition method is not limited to this. For example, the center of edge feature points at a fixed interval or a preset position may be used as the sampling point.

  In step 204 of this example, the visual features include a saturation feature, a spacing feature, and a luminance difference feature. Among them, the saturation feature may be a gray level feature of the selected sampling point in the detection waiting image, and the spacing feature is a spacing feature between the selected adjacent edge feature points in the detection waiting image. Alternatively, the luminance difference feature may be a difference feature due to a gray level between adjacent sampling points in the detection-waiting image. Note that the present embodiment is not limited to this. For example, the saturation feature may be the RGB value feature of the sampling point of the selected row in the detection waiting image, and the interval feature is in the detection waiting image. For example, a spacing feature between selected adjacent sampling points.

  Although how to obtain visual features has been described above by way of example, how to detect zebra crossing based on visual features will be described below.

  In step 102 in this embodiment, it is determined whether or not zebra crossing is present in the image based on a statistical result for the visual feature, and the statistics include a saturation feature, a spacing feature, and a luminance difference feature. Includes statistics for each.

  In an example of step 102, statistics are performed on the visual characteristics of the selected row, the saturation features are alternately black and white, the spacing features are equally spaced, and the luminance difference features are alternately continuous. The sampling point that coincides (matches) with the light and dark to be determined is a feature point of zebra crossing, and the number of feature points of the zebra crossing that are continuous at the maximum is obtained by statistics to be a first statistical value. When the value is greater than the third threshold, it is determined that zebra crossing is present in the image.

  Among them, as a method of judging that the saturation feature is alternately black and white, among the gray levels of the three adjacent sampling points, the gray level of the intermediate sampling point is simultaneously the adjacent sampling points before and after that. There is a method for determining that the saturation feature is alternately black and white when the gray level is smaller or larger than the gray level.

  For example, if the gray levels of three adjacent sampling points are 10, 250, 15 or 240, 15, 245, respectively, it is determined that the saturation feature is alternating black and white, or the RGB values of the sampling points May be calculated directly, and the RGB values of three adjacent sampling points are (255, 255, 255), (0, 0, 0), (255, 255, 255) or (0, 0, 0), respectively. , (255, 255, 255), (0, 0, 0), it is determined that the saturation feature is black and white alternately continuous. Here, an example has been described in which the RGB value is pure white or pure black. However, when actually processing, the RGB values may be set as black and white thresholds according to needs. Therefore, detailed description is omitted here.

  Among them, as a method of determining that the interval feature is equally spaced, there is a method of determining that the interval feature is equally spaced when the ratio of the interval between the adjacent sampling points does not exceed the second threshold, Here, when calculating the ratio of the intervals between adjacent sampling points, it may be set in advance that a ratio between a relatively large interval value and a relatively small interval value among the intervals between adjacent sampling points is used. Good.

  For example, here, the interval between adjacent sampling points may be calculated based on the coordinate values of the sampling points in the image. For example, the coordinates of three adjacent sampling points are (10, 10) and (25 , 10), (39, 10), the interval between adjacent sampling points is 15 and 14, respectively, the ratio of the intervals is 15/14, and if the second threshold is 1.5, the interval feature is Judged at equal intervals. Alternatively, when calculating the ratio of the intervals between adjacent sampling points, it may be preset to use a ratio between a relatively small interval value and a relatively large interval value among the intervals between adjacent sampling points. When the ratio is greater than a predetermined threshold, it is determined that the interval feature is equally spaced, or when the absolute value of the difference between the intervals of adjacent sampling points by calculation does not exceed the second threshold, May be determined to be equally spaced, for example, according to the calculation as described above, adjacent sampling point intervals are 15 and 14, respectively, the absolute value of the difference between the intervals is 1, and the second threshold value is If it is 2, it may be determined that the interval features are equally spaced. The present invention is not limited to this, and for example, by calculating the ratio or difference between the intervals of adjacent edge feature points, it may be determined that the interval features are equally spaced. Is similar to the determination method based on the sampling point, and thus detailed description thereof is omitted here.

  Among them, as a method for determining whether the brightness difference feature is alternately continuous light and dark, when the absolute value of the gray level difference between two adjacent sampling points is larger than the first threshold, the brightness difference feature is alternately continuous. There is a method to judge that it is light and dark. For example, if the gray levels of two adjacent sampling points are 10 and 250, respectively, the absolute value of the difference between the gray levels is 240, and when the first threshold is 200, the luminance difference features are alternately continuous. Judged to be light and dark. When actually processing, the first threshold value can be arbitrarily set according to the needs, and the present embodiment is not limited to this, and it is determined whether or not the luminance difference feature is alternately continuous light and dark. Therefore, for example, the first threshold value can be set to a relatively large value.

  In the above, the determination method has been described by way of example only in which the saturation feature is alternately black and white, the interval feature is equally spaced, and the brightness difference feature is alternately continuous light and dark. However, the present invention is not limited to this. For example, the saturation feature may be black-and-white that is alternately continuous, the interval feature may be approximately equal intervals, and the luminance difference feature may be light and dark that are substantially alternately alternating, depending on the actual situation. Appropriate adjustments may be made to determine specific implementation methods.

  In this example, after determining whether the saturation feature is alternately black and white, whether the interval feature is equally spaced, and whether the luminance difference feature is alternately continuous light / dark, the above three types are used. Sampling points that match (ie, all the judgment results are “Yes”) are determined as zebra crossing feature points, and the number of the zebra crossing feature points that are the largest consecutive by statistics (calculation) It is obtained as a statistical value, and when the first statistical value is larger than the third threshold value, it is determined that zebra crossing is present in the image.

  For example, there are a total of 20 sampling points in the selected row, of which the number of sampling points matching the above three cases is 15, that is, 15 zebra crossing feature points. Yes, of which the maximum number of consecutive zebra crossing feature points is 11, that is, among 11 consecutive sampling points (zebra crossing feature points), the visual feature is alternating with the saturation feature Are all black and white, the interval features are equally spaced, and the luminance difference features are alternately continuous light and dark. At this time, if the third threshold value is 10, it may be determined that zebra crossing exists in the image. Among them, the third threshold value can be arbitrarily set according to actual needs.For example, the third threshold value may be 80% of the number of edge feature points. Not limited.

  The above has been described by taking only one line of the image as an example. In another example of this embodiment, in order to further improve the accuracy of zebra crossing detection, in step 101, multiple rows of data or all rows (all rows) of data in the image are selected, and , The visual characteristics of the selected multiple rows of data or the entire rows of data may be respectively calculated, of which the selected multiple rows of data may be any multiple rows of data in the image, The data may be a plurality of rows selected according to a predetermined rule. Of these, the visual feature calculation method can be referred to the above-described example, and thus detailed description thereof is omitted here.

  In step 102, based on the selected plurality of lines, the statistics of the first statistical value are performed, and the second statistical value is calculated. When the second statistical value is greater than the fourth threshold, it is determined that zebra crossing is present in the image. Among them, the second statistical value may be a maximum value or an average value of the first statistical values by statistics for each of the selected plurality of rows.

  For example, if 5 rows of data are selected and the first statistical value obtained by statistics is 11, 10, 12, 11, 11, the second statistical value is 12 (maximum value) or 11 (average value) ). If the fourth threshold is 10, it can be determined that zebra crossing is present in the image.

  Among them, the calculation of the second statistical value is not limited to the calculation of the maximum value or the average value of the plurality of first statistical values, and the selection of the number of rows and the fourth threshold value are arbitrarily set according to the actual needs. In addition, the fourth threshold value may be the same value as the third threshold value or may be a different value, and the present embodiment is not limited to this.

  As can be seen from the above-described method of the embodiment of the present invention, visual features in the image, that is, visual features including a saturation feature, a spacing feature, and a luminance difference feature are calculated simultaneously, and based on a statistical result of the visual features. Thus, by determining whether zebra crossing is present in the image, the efficiency and effectiveness of immediate detection of zebra crossing can be improved, and false detection of zebra crossing can be avoided.

  The second embodiment of the present invention further provides a method for detecting zebra crossing in an image, and the detection method will be described based on the first embodiment. Moreover, although this method can be used for a driving assistance system, this invention is not limited to this.

  FIG. 3 is a flowchart of a method for detecting zebra crossing in an image according to the second embodiment of the present invention, which is only an example, and for example, in FIG. 3, some steps may be omitted. Further, as described above, some steps may be replaced by other implementation methods, and thus detailed description thereof is omitted here. As shown in FIG. 3, the method includes the following steps.

  Step 301: Perform pre-processing on the image.

  Among them, in step 301, the image may be a forward road image acquired immediately by a camera installed in the front window of the vehicle, and is acquired by extracting an image frame in the road immediate video. It may be. The pre-processing for the image is mainly performed by normalizing the image waiting for detection to a standard size, and performing noise removal processing (for example, by using a Gaussian smoothing filter) on the normalized image to obtain noise. It may also include converting the image after removal into a grayscale image.

Step 302: Obtain edge features in a grayscale image;
Step 303: Obtain a region of interest of the image.

  FIG. 4 is a diagram illustrating the region of interest acquired in step 303 of the present embodiment. As shown in FIG. 4, in the image 401, a region of interest 402 having coordinates (X0, Y0) as start point coordinates, a width of W, and a height of H may be acquired. By acquiring the region of interest, interference due to invalid information in the image can be reduced, and the efficiency of image processing can be improved.

  Step 304: Initialize parameters. Of these, for example, the following parameters may be initialized.

  Statistical value of the number of rows RowCnt = Y0; Initial value of the first statistical value RowMaxVote = 1; Initial value of the second statistical value may be initialized as FinalMaxVote = 1, of which the first statistical value and the second statistical value Since the definition of is the same as that of the first embodiment, detailed description thereof is omitted here.

Step 305: Select one row of data in the region of interest and set its row number y = RowCnt;
Step 306: The edge feature point Xe (n) in the selected one line of data is determined.

  FIG. 5 is a diagram showing edge feature points in the edge feature image in step 306 of the present embodiment. For example, the following edge feature points, that is, Xe1, Xe2, Xe3, Xe4,... Xe9,... May be determined, and among them, a high feature value, for example, 255 is assigned (given), and a non-edge feature point A low feature value, for example, 0 may be assigned. If the edge feature in the selected row is an edge line segment, the starting point of the line segment may be selected as the edge feature point, for example, Xe9. However, the present invention is not limited to this.

  Step 307: Determine sampling points based on edge feature points.

  FIG. 6 is a diagram showing sampling points in step 307 of the present embodiment. Of these, the center of two adjacent edge feature points in the plurality of edge feature points may be used as one sampling point of the selected row, and thus one or more sampling points may be expressed as Can be obtained.

Xs (n) = (Xe (n) + Xe (n + 1)) / 2 (1)
As shown in FIG. 6, the following sampling points, that is, Xs1, Xs2, Xs3, Xs4,..., Xs8,.

  Step 308: Calculate visual features of the selected row in the image based on the plurality of edge feature points and the one or more sampling points.

  In the present embodiment, the visual feature of the selected row may be calculated by the following method.

Brightness difference feature DG (n) = ABS [G (n) −G (n−1)] (2)
Saturation feature BW (n) = (G (n)> G (n−1))? 1: -1 (3)
Spacing feature RL (n) = L (n) / L (n−1) = (Xs (n + 1) −Xs (n)) / (Xs (n) −Xs (n−1)) (4)
G (n) represents the gray level of the sampling point Xs (n), ABS [x] represents the absolute value of the variable x, and L (n) is the interval between two adjacent sampling points. Represents Note that the above is only an exemplary visual feature calculation method, and the present embodiment is not limited to this.

  Step 309: Perform statistics on the visual characteristics of the selected line, and the saturation feature is alternately black and white, the interval feature is equally spaced, and the brightness difference feature is alternately light and dark. Sampling points that coincide with each other are determined as zebra crossing feature points, and the number of the maximum zebra crossing feature points is obtained as a first statistical value.

  In this embodiment, the following is an example of a code for calculating the first statistical value in step 309.

LOOP (n = 1 to k)
{
TempVoteCnt = (ZFPDecision [DG (n), BW (n), RL (n)] = true)? (TempVoteCnt + 1): 0;
RowMaxVote = (TempVoteCnt ≧ RowMaxVote)? TempVoteCnt: RowMaxVote;
}
Among them, TempVoteCnt is a temporary variable of the first statistic value, and ZFPDecision [DG (n), BW (n), RL (n)] is black and white in which the saturation feature in the visual feature continues alternately Represents a determination function for determining whether the interval feature is equally spaced or whether the luminance difference feature is alternately continuous light and dark, and the implementation method of the function is the same as step 102 in Example 1. Since they are similar, detailed description is omitted here.

  Step 310: After calculating the first statistical value, it is determined whether or not the processing for all the selected rows of data in the image has been completed. If the determination result is “No”, Step 305 is performed. Otherwise, perform step 311.

  In this embodiment, when RowCnt> ROW_NUM_END, it is determined that step 310 is performed. ROW_NUM_END represents the row number of the last row of the region of interest, that is, ROW_NUM_END == Y0 + H−1, otherwise The loop condition is set, that is, RowCnt = RowCnt + STEP, where STEP is a preset row number constant between the selected rows, and Step 305 is performed after updating the RowCnt value.

  Step 311: Calculate a second statistic value based on the first statistic value according to the statistics performed for each of the selected plurality of lines, and when the second statistic value is greater than a fourth threshold value, It is determined that zebra crossing exists.

  In the present embodiment, the second statistical value is the maximum value of the first statistical value according to the statistics of the selected plurality of lines, and the following is a code for calculating the second statistical value in step 311: It is an example.

if (RowMaxVote ≧ FinalMaxVote)
{
FinalMaxVote = RowMaxVote;
}
Among them, the calculation method of the second statistical value is not limited to this, and the calculation method in the first embodiment can also be referred to, and thus detailed description thereof is omitted here.

  Also shown below is an example of code that determines whether zebra crossing is present at step 311.

if (FinalMaxVote> ZFP_NUM_TH)
Zebra-Crossing exists;
else
Zebra-Crossing not exist.
Among them, ZFP_NUM_TH is the fourth threshold value, and its setting method is similar to that of the first embodiment, and therefore detailed description thereof is omitted here.

  As can be seen from the above-described method of the embodiment of the present invention, visual features in the image, that is, visual features including a saturation feature, a spacing feature, and a luminance difference feature are calculated simultaneously, and based on a statistical result of the visual features. Thus, by determining whether or not zebra crossing is present in the image, the speed and effectiveness of immediate detection of zebra crossing can be improved, and false detection of zebra crossing can be avoided.

  Embodiment 3 of the present invention further provides a device for detecting zebra crossings in an image, and the principle by which the device solves the problem is similar to the method of Embodiments 1 and 2, and therefore, its specific implementation. Since the implementation of Example 1 or Example 2 can be referred to, description of the same contents is omitted here.

  FIG. 7 is a configuration diagram of a zebra crossing detecting device 777 in an image according to the third embodiment of the present invention. As shown in FIG. 7, the apparatus 777 includes a first calculation unit 701 and a first processing unit 702.

A first calculation unit 701: calculating visual features of the image based on the detected edge features of the image, wherein the visual features include a saturation feature, a spacing feature and a luminance difference feature;
First processing unit 702: Based on the statistical result of the visual feature calculated by the first calculation unit 701, it is determined whether or not zebra crossing exists in the image.

  In this embodiment, in order to improve detection efficiency, the device 777 may further include a second processing unit 700, which performs pre-processing on the image.

  FIG. 8 is a configuration diagram of the first calculation unit 701 in the embodiment of the present invention, and the first calculation unit 701 includes the following.

First selection unit 801: selecting one row of data in the image;
First detection unit 802: detecting a plurality of edge feature points of the selected row of data;
First acquisition unit 803: The center of two adjacent edge feature points among the plurality of edge feature points detected by the first detection unit 802 is set as one sampling point in a selected row, thereby one Or obtain multiple sampling points;
Second calculation unit 804: The visual feature of the image is calculated based on the plurality of edge feature points detected by the first detection unit 802 and one or more sampling points acquired by the first acquisition unit 803.

  FIG. 9 is a configuration diagram of the first processing unit 702 in the embodiment of the present invention, and the first processing unit 702 includes the following.

First statistical unit 901: Performs statistics on the visual characteristics of the selected row, and is black and white in which the saturation feature is alternately continuous, the interval feature is equally spaced, and the brightness difference feature is alternately continuous. A sampling point that coincides with a zebra crossing feature point is determined, and the number of maximum continuous zebra crossing feature points is obtained by statistics to be a first statistical value;
First determination unit 902: When the first statistical value according to the statistics of the first statistical unit 901 is larger than the third threshold, it is determined that zebra crossing is present in the image.

  In this embodiment, the first statistical unit 901 further includes:

First determination unit 9011: When the absolute value of the difference between the gray levels of two adjacent sampling points acquired by the first acquisition unit 803 is larger than the first threshold value, the brightness difference feature is alternately light and dark. Judge;
Second determination unit 9012: Among the gray levels of three adjacent sampling points acquired by the first acquisition unit 803, the gray level of the intermediate sampling point is simultaneously smaller than the gray levels of the adjacent sampling points before or after that, or When large, determine that the saturation feature is alternating black and white;
Third determination unit 9013: When the ratio of the intervals between adjacent sampling points detected by the first detection unit 802 does not exceed the second threshold, it is determined that the interval features are equally spaced.

  Among them, since the specific implementation method of the first statistical unit 901, the first determination unit, the first determination unit 9011, the second determination unit 9012, and the third determination unit 9013 is similar to the first embodiment, here, Description of the contents is omitted.

  In another implementation method of this embodiment, in order to further improve the accuracy of zebra crossing detection, the first selection unit 801 further selects data of all lines of the image or data of some lines. Alternatively, the first statistical unit 902 may further obtain a plurality of first statistical values of the data in the row selected by the first selection unit 801, respectively.

  The first processing unit 702 may further include:

Second statistical unit 903: The first statistical unit 901 calculates a second statistical value based on the plurality of first statistical values obtained by the statistics, and the second statistical value is the maximum of the plurality of first statistical values. Value or average value;
Second deterministic unit 904: When the second statistical value calculated by the second statistical unit 903 is larger than the fourth threshold value, it is determined that zebra crossing is present in the image.

  Embodiments of the present invention further provide an apparatus for detecting zebra crossings in an image.

  FIG. 10 is a configuration diagram of a device for detecting zebra crossing in an image according to an embodiment of the present invention. As shown in FIG. 10, the detection device 1000 may include a central processor (CPU) 200 and a storage 210, which is coupled to the central processor 200. Among them, the storage device 210 may store various data, for example, visual characteristics and statistical results of the image, and also stores a program for detecting zebra crossing in the image, and the central processing. The program is executed under the control of the vessel 200 to determine whether zebra crossing is present in the image.

  In one example, the function of a device for detecting zebra crossings in the image described above may be integrated in the central processor 200. Among them, the central processor 200 calculates the visual features of the image based on the detected edge features of the image, and the visual features include a saturation feature, a spacing feature, and a luminance difference feature; and statistics on the visual features It may be configured to determine whether zebra crossings are present in the image based on the results.

  Among them, calculating the visual feature of the image based on the detected edge feature of the image selects one row of data in the image; detects a plurality of edge feature points of the data of the selected row; One or more sampling points are obtained by taking the center of two adjacent edge feature points in the edge feature points as one sampling point of the selected row; multiple edge feature points and one Alternatively, calculating visual features of selected rows in the image based on the plurality of sampling points.

  Of these, determining whether zebra crossings are present in an image based on the statistical results for visual features is based on the visual characteristics of the selected row, and the saturation features are alternately repeated. The sampling point that corresponds to black and white, the interval feature is equally spaced, and the brightness difference feature is alternately continuous light and dark is determined as the zebra crossing feature point, and the maximum continuous zebra crossing Including obtaining the number of feature points as a first statistical value; and determining that zebra crossing is present in the image when the first statistical value is greater than a third threshold.

  Among them, the central processor 200 may be further configured as follows.

  When the absolute value of the difference between the gray levels of two adjacent sampling points is greater than the first threshold, it is determined that the luminance difference feature is alternately continuous light and dark; When the gray level of the sampling points of the current sampling point is smaller or larger than the gray level of the adjacent sampling points before and after the sampling point, it is determined that the saturation feature is alternately black and white; When the second threshold is not exceeded, it is determined that the interval features are equally spaced.

  Alternatively, perform statistics on the visual characteristics of all lines of the image, or select some lines and perform statistics on the visual characteristics; first by statistics for each of a plurality of selected lines. Calculating a second statistic based on the statistic; determining that a zebra crossing is present in the image when the second statistic is greater than a fourth threshold; wherein the second statistic is a plurality of selected It may be the maximum value or the average value of the first statistical values according to the statistics of each row.

  In another example, the zebra crossing detection device in the image may be arranged separately from the central processor, for example, the zebra crossing detection device in the image is arranged as a chip connected to the central processor 200. The function of the zebra crossing detecting device in the image may be realized by the control of the central processor.

  Also, as shown in FIG. 10, the zebra crossing detection device in the image may further include a transmitter / receiver 220, a sensor 230, etc., of which the functions of the above-mentioned components are similar to those of the prior art, and are detailed here. Description is omitted. Note that the zebra crossing detecting device 1000 in the image does not necessarily include all the components shown in FIG. Further, the zebra crossing detection device 1000 in the image may further include other parts not shown in FIG. 10, and the prior art can be referred to for this.

  As can be seen from the above-described apparatus of an embodiment of the present invention, visual features in the image, i.e. visual features including a saturation feature, a spacing feature, and a luminance difference feature are simultaneously calculated and Based on this, by determining whether zebra crossing is present in the image, the speed and effectiveness of immediate detection of zebra crossing can be improved, and false detection of zebra crossing can be avoided.

  Embodiments of the present invention further provide a computer readable program, wherein when executing the program in a detection device for zebra crossings in an image, the program is stored in a detection device for zebra crossings in the image. The method for detecting zebra crossing in an image described in the first embodiment or the second embodiment is executed.

  An embodiment of the present invention further provides a storage medium storing a computer readable program, wherein the computer readable program is stored in a zebra crossing detection device in an image described in the first embodiment or the second embodiment. A method for detecting zebra crossings in an image is executed.

  Note that the above-described apparatus and method of the present invention may be realized by hardware, or may be realized by a combination of hardware and software. The present invention also relates to such a computer-readable program, that is, when the program is executed by the logic unit, the logic unit can cause the above-described apparatus or component to be realized. Alternatively, the above-described various methods or steps can be realized in the logic unit. The present invention further relates to a storage medium storing the above-mentioned program, for example, a hard disk, a magnetic disk, an optical disk, a DVD, a flash memory, and the like.

  Moreover, the following additional remarks are disclosed regarding the embodiment including each of the above-described examples.

(Appendix 1)
A device for detecting zebra crossings in an image,
A first calculation unit for calculating visual features of the image based on detected edge features of the image, wherein the visual features include a saturation feature, a spacing feature, and a luminance difference feature; and An apparatus comprising: a first processing unit that determines whether zebra crossings are present in the image based on statistical results for the visual features calculated by the first calculation unit.

(Appendix 2)
The apparatus according to appendix 1, wherein
The first calculation unit is:
A first selection unit for selecting one row of data in the image;
A first detection unit for detecting a plurality of edge feature points in a selected row of data;
One or a plurality of sampling points by setting the center of two adjacent edge feature points among the plurality of edge feature points detected by the first detection unit as one sampling point in a selected row A first acquisition unit that obtains the image visual feature based on the plurality of edge feature points detected by the first detection unit and the one or more sampling points acquired by the first acquisition unit. A device comprising a second calculation unit for calculating.

(Appendix 3)
The apparatus according to appendix 2, wherein
The first processing unit is
Statistics are performed on the selected visual characteristic of the selected line, and the saturation feature is black and white alternately, the interval feature is equally spaced, and the brightness difference feature is light and dark alternately. A first statistical unit that determines a sampling point that coincides with a zebra crossing feature point as a first statistical value by statistically acquiring the number of the maximum zebra crossing feature points; and the first statistical unit; An apparatus comprising: a first deterministic unit that determines that zebra crossing is present in the image when the first statistical value according to the statistic is greater than a third threshold.

(Appendix 4)
The apparatus according to appendix 3,
The first statistical unit is
When the absolute value of the difference between the gray levels of two adjacent sampling points acquired by the first acquisition unit is greater than a first threshold, it is determined that the brightness difference feature is alternately continuous light and dark Judgment unit;
When the gray level of the intermediate sampling point among the gray levels of the three adjacent sampling points acquired by the first acquisition unit is smaller or larger than the gray level of the adjacent sampling points before and after the sampling point, A second determination unit for determining that the degree feature is alternately black and white; and the interval feature when the ratio of the intervals between the adjacent sampling points detected by the first detection unit does not exceed a second threshold value. Including a third determination unit that determines that are equally spaced.

(Appendix 5)
The apparatus according to appendix 4, wherein
The first selection unit further selects data of all lines or a part of lines of the image,
Each of the first statistical units further obtains a plurality of first statistical values of a plurality of rows of data selected by the first selection unit,
Of these, the first processing unit further comprises:
A second statistical unit for calculating a second statistical value based on the first statistical value by statistics for each of the plurality of selected rows, wherein the second statistical value is the plurality of first statistical values; A second statistical unit that is a maximum or average value; and a second statistical unit that determines that zebra crossing is present in the image when the second statistical value calculated by the second statistical unit is greater than a fourth threshold. A device containing two deterministic units.

(Appendix 6)
The apparatus according to appendix 1, wherein
An apparatus further comprising a second processing unit for performing a pre-processing on the image.

(Appendix 7)
A method for detecting zebra crossings in an image, comprising:
Calculating visual features of the image based on the detected edge features of the image, the visual features including a saturation feature, a spacing feature, and a luminance difference feature; and based on a statistical result for the visual feature Determining whether zebra crossings are present in the image.

(Appendix 8)
The method according to appendix 7,
Calculating the visual features of the image based on the detected edge features of the image;
Selecting a row of data in the image;
Detecting a plurality of edge feature points in a selected row of data;
Obtaining one or more sampling points by taking the center of two adjacent edge feature points in the plurality of edge feature points as one sampling point in a selected row; and Calculating a visual feature of a selected row of the image based on an edge feature point and the one or more sampling points.

(Appendix 9)
The method according to appendix 8, wherein
Calculating a selected line of visual features in the image comprises:
The luminance difference feature DG (n) = ABS [G (n) −G (n−1)];
The saturation feature BW (n) = (G (n)> G (n−1))? 1: −1; and the interval feature RL (n) = L (n) / L (n−1) = (Xs (n + 1) −Xs (n)) / (Xs (n) −Xs (n−1) )
Among them, G (n) indicates a gray level of a sampling point, and L (n) indicates a distance between adjacent sampling points Xe (n) and Xe (n + 1).

(Appendix 10)
The method according to appendix 8, wherein
Determining whether zebra crossings are present in the image based on statistical results for the visual features,
Statistics are performed on the visual characteristics of the selected row, and the saturation feature is alternately black and white, the interval feature is equally spaced, and the brightness difference feature is alternately continuous light and dark. Sampling points that match the above are determined as zebra crossing feature points, and the number of the maximum continuous zebra crossing feature points is statistically obtained as a first statistical value; and Determining that there is zebra crossing in the image when greater than a threshold.

(Appendix 11)
The method according to appendix 10, wherein
When the absolute value of the difference in gray level between two adjacent sampling points is greater than a first threshold, it is determined that the brightness difference feature is alternately continuous light and dark;
When the gray level of the intermediate sampling point among the gray levels of the three adjacent sampling points is simultaneously smaller or larger than the gray level of the adjacent sampling points before and after that, the black and white in which the saturation feature is alternately continuous Further comprising: determining that the spacing feature is equidistant when a ratio of spacing between adjacent sampling points does not exceed a second threshold.

(Appendix 12)
The method according to appendix 11, wherein
Perform statistics on visual characteristics of all lines of the image, or select some lines and perform statistics on the visual characteristics;
Calculating a second statistic based on the first statistic with statistics for each of a plurality of selected rows; and when the second statistic is greater than a fourth threshold, in the image Further comprising determining that zebra crossing is present;
Wherein, the second statistical value is a maximum value or an average value of the first statistical values according to statistics for each of a plurality of selected rows.

(Appendix 13)
The method according to appendix 7,
A method further comprising performing a pre-processing on the detection waiting image before detecting edge feature data of the image.

  The preferred embodiment of the present invention has been described above, but the present invention is not limited to this embodiment, and all modifications to the present invention belong to the technical scope of the present invention unless departing from the spirit of the present invention.

Claims (8)

A device for detecting zebra crossings in an image,
A first calculation unit for calculating visual features of the image based on detected edge features of the image, wherein the visual features include a saturation feature, a spacing feature and a luminance difference feature; And a first processing unit for determining whether zebra crossing is present in the image based on a statistical result for the visual feature calculated by the first calculation unit ,
The first calculation unit is:
A first selection unit for selecting one row of data in the image;
A first detection unit for detecting a plurality of edge feature points of the selected row of data;
One or more samplings by setting the center of two adjacent edge feature points among the plurality of edge feature points detected by the first detection unit as one sampling point of the selected row. A first acquisition unit for acquiring points; and
A second calculation unit for calculating the image visual feature based on the plurality of edge feature points detected by the first detection unit and the one or more sampling points acquired by the first acquisition unit; , Equipment. The apparatus of claim 1 , comprising:
The first processing unit is
Statistics are performed on the selected visual characteristics of the one row, and the saturation features are alternately black and white, the interval features are equally spaced, and the brightness difference features are alternately continuous. A first statistical unit that determines that a sampling point that coincides with a zebra crossing feature point is a zebra crossing feature point, obtains a maximum number of consecutive zebra crossing feature points by statistics, and sets a first statistical value; An apparatus comprising: a first deterministic unit that determines that zebra crossing is present in the image when the first statistical value according to the statistics of one statistical unit is greater than a third threshold. The apparatus of claim 2 , comprising:
The first statistical unit is
When the absolute value of the difference between the gray levels of two adjacent sampling points acquired by the first acquisition unit is greater than a first threshold, it is determined that the brightness difference feature is alternately continuous light and dark Judgment unit;
When the gray level of the intermediate sampling point among the gray levels of the three adjacent sampling points acquired by the first acquisition unit is smaller or larger than the gray level of the adjacent sampling points before and after the sampling point, A second determination unit for determining that the degree feature is alternately black and white; and the interval feature when the ratio of the intervals between the adjacent sampling points detected by the first detection unit does not exceed a second threshold value. Including a third determination unit that determines that are equally spaced. The apparatus of claim 3 , comprising:
The first selection unit further selects data of all lines or a part of lines of the image,
Each of the first statistical units further obtains a plurality of first statistical values of a plurality of rows of data selected by the first selection unit,
The first processing unit is a second statistical unit that calculates a second statistical value based on the first statistical value based on statistics for each of the selected plurality of rows, and the second statistical value Is a second statistical unit that is a maximum or average value of the plurality of first statistical values; and the second statistical value calculated by the second statistical unit is greater than a fourth threshold value in the image. An apparatus comprising a second deterministic unit that determines that zebra crossing is present. A method for detecting zebra crossings in an image, comprising:
Calculating visual features of the image based on detected edge features of the image, the visual features including a saturation feature, a spacing feature, and a luminance difference feature; and based on statistical results for the visual feature Determining whether zebra crossing is present in the image ,
Calculating visual features of the image based on detected edge features of the image,
Selecting a row of data in the image;
Detecting a plurality of edge feature points of the selected row of data;
Obtaining one or more sampling points by taking the center of two adjacent edge feature points in the plurality of edge feature points as one sampling point of the selected row; and
Calculating the selected row of visual features in the image based on the plurality of edge feature points and the one or more sampling points . 6. A method according to claim 5 , wherein
Determining whether zebra crossings are present in the image based on statistical results for the visual features,
Statistics are performed on the selected visual characteristics of the one row, and the saturation features are alternately black and white, the interval features are equally spaced, and the brightness difference features are alternately continuous. A sampling point that coincides with the existence of a zebra crossing feature point is determined as a zebra crossing feature point, and the number of the maximum zebra crossing feature points is obtained from the statistics as a first statistical value; and Determining that there is zebra crossing in the image when greater than a third threshold. The method of claim 6 , comprising:
When the absolute value of the difference in gray level between two adjacent sampling points is greater than a first threshold, it is determined that the brightness difference feature is alternately continuous light and dark;
When the gray level of the intermediate sampling point among the gray levels of the three adjacent sampling points is simultaneously smaller or larger than the gray level of the adjacent sampling points before and after that, the black and white in which the saturation feature is alternately continuous Further comprising: determining that the spacing feature is equidistant when a ratio of spacing between adjacent sampling points does not exceed a second threshold. The method of claim 7 , comprising:
Perform statistics on visual characteristics of all lines of the image, or select some lines and perform statistics on the visual characteristics;
Calculating a second statistical value based on a first statistical value for each of the selected plurality of rows; and when the second statistical value is greater than a fourth threshold, Further comprising determining that the crossing is present;
The method, wherein the second statistical value is a maximum value or an average value of first statistical values according to statistics for each of the selected plurality of rows.

Images