TW201040847A - Method of identifying objects in image - Google Patents

Abstract

A method of identifying objects in images includes steps of: photographing an object; capturing characteristics of the object to generate histogram of the X-axis or the Y-axis of the captured object; deriving a minimum specific point and a maximum characteristic point from the histogram of the X-axis or the Y-axis; calculating the shape, size, and position information of the object by polynomial regression analysis based on the information of minimum specific point and the maximum characteristic point; if the object is identified to be linear trend, the object is determined to be an triangle; if the object is identified to be straight line trend, the object is determined to be a square; and if the object is identified to comply with a quadratic equation, the object is determined to be a circle.

Description

201040847 VI. Description of the Invention: [Technical Field] The present invention relates to a method for recognizing an object in an image, in particular to an image that can accurately and quickly recognize the shape, size or position of the captured image. The method of the object. [Prior Art] A conventional object recognition system includes a camera (or the like) for capturing an input image, and includes cutting an input image, extracting an object from the image of the cut image, and representing the image object of the input image. And will be identified 撷: Object classification. The input image is composed of a digital format <color or gray scale pixel data and is arranged in a two-dimensional array of gamma directions, wherein the image may contain thousands or tens of thousands of independent pixels. At the lowest level of the image representation, the object can be modeled as an entire image, and the image is compared to the original input image in the database by - pixels to pixels. However, most object recognition systems use different cut images and methods of capturing, representing, and classifying image objects to improve speed and/or accuracy (Kmmm, U.S. Patent 7,092,566). In particular, the color strip chart has been used for a variety of different processing steps, which are primarily used to increase the speed of object recognition in color images. In U.S. Patent No. 7, 〇 2, 329, Prempraneerach et al. describe a technique for cutting a color image into a plurality of regions using a color strip chart by converting the image into a three-dimensional color space. Generating a color strip chart for each dimension of the color space, using the 201040847 long gauge map to generate a plurality of junction boxes in the three-dimensional color space to calculate a normalized variation value of each of the connected boxes to form a flail box The cluster group, corresponding to the clustered pixel image of the image region that is cut and corresponding to the color characteristic of the phase, to take the cut region from the image, and classify the grouped pixels in the image domain to identify the image The object in it. For the classification method, you can use the neural network (with adaptive template matching technology, frequency-sensitive competition learning, 'opportunity penalty competition learning or statistical classification and classification method. Ο Ο In the past, rrrneerach et al. explained that the proposed clustering technique is better. = De-stacking clustering technique is more effective, and it is also pointed out that the proposed method of cutting d is more intensively per pixel-only method to establish the strength of the long-statistic chart required for post-cluster grouping. H and saturation. π and Bard However, this method is a complex calculation for the object recognition processing. First, the input of the silly must, by: still need to be connected with h Μ β image (4) - edge reservation filter to deal with To smooth the color image, the discontinuity in the long chart, and then the image = the long slab of the chart s, which is expected to be in the long-range data. The lowest point is a long chart. ..., the mirror is wrapped around the mother. The connected box group is formed by calculating the difference in pixel values in each of the junction boxes, and the junction box is constructed according to the normalized variation value. The minimum normalized variation value of the area of the connection box = <, the root node of the 'female structure', and the others are the most steep ladder. The 201040847 degree reduction algorithm is linked to the branch node of the root node. In order for this method to work properly, the same f-region in the image must be clustered into a well-defined spacing in the three-dimensional strip chart, but the true image color variation may be difficult to convert to noise or RGB to ms color space. The color variation phase formed by the nonlinear transformation is separated. Ο

In U.S. Patent Nos. 7,092,566, 6,952,496 & 6,611,622, Krumm describes the use of color strip charts to represent and classify an input image technique. The object_method first establishes and stores a model strip chart of the object to be identified, and then (4)-wheel image to take out the area that may be waiting for the object to be identified, and obtain the long purity from the cut area. Then, the long graph and the stored model strip graph are obtained. If the similarity of the two long bars exceeds a predetermined threshold, it means that the input image matches the model object. The matching input strip chart can also be added to the database of the model's long strip chart of the fixed object. The above method establishes a model residual image strip chart by determining the true RGB color represented by the pixel in a model or input image region, and the full range of the true pixel color (four) into a series of discrete color ranges or quantized colors. The category 'assigns the extracted model or each pixel of the input image area to the quantized color category and establishes a count value assigned to the number of pixels per 'quantized color category'. In a preferred embodiment, the decimated pixel values are quantized into 27 color categories by comparing the pixel count values in each quantized color category to the input scene; the comparison of the image and the model strip chart. The model strip chart must be obtained from the prefat〇ry image t similar to the input object to be identified. The image area of the model strip chart is also used in the subsequent input image, 201040847 to be taken out. The object to be identified. The self-phased 'model image is cut by the following method: pixel value determination without significant change - static f scene image, the background image is extracted by the image - foreground image 2 changes the pixel value group to cut the foreground image g object Ο Ο However, this method is mainly used to track the objects in the time series image. In addition, the method still needs a large amount of modeling time. The longer the statistical chart is, the more processing time is than the input image long chart and the model long chart. The color strip chart generation technique used in the method does not reserve the original geometry as a summary of the number of image pixels of a fixed color, so the shape, size and position of the real object cannot be determined. Finally, similar strip graphs of different objects may result in incorrect object recognition results. In U.S. Patent Nos. 6,532,301 and 6,477,272, Kappa describes the use of co-occurrence strip charts to represent and confirm the location of a modeled object in a search image. (10) The above method first establishes a model image of the object, and then calculates a co-growth bar graph for each model image. The image of the model is created by the isometric angle of the circumference of the object, but the image points of the object to be confirmed are captured by the viewpoints of different distances, and the calculation of the total growth bar chart is confirmed by confirming the image of the model. Each of the possible, and unique, unaligned pairs of pixels and the number of pairs of pixels that fall within the same color range. Then, a search window of a predetermined size is generated from the overlay portion of the search image, and a common growth bar graph of each search window is a pixel color and a space established by using the technology and the model image co-growth bar graph. Produced from the range. 201040847 Finally, each model image co-growth bar chart is compared with each search window co-growth bar chart to evaluate its similarity. Each model image co-growth bar graph matching search window co-growth bar graph is set to possibly include the object to be identified, wherein the matching system is performed by a similar value of a large value = a critical value. Then, the position of the object to be identified is determined to be located in a single search area having the largest similarity measure among all the search windows that may include the object to be recognized, and may be repeated upward, downward, left or right. Moving a pixel position, then calculating the search window total growth bar statistic®, and comparing the search window co-growth bar graph with each model strip graph to obtain a possibly higher similarity measure value. The system and method must have its search window size, color range, and range selected at the beginning of the image search.

Knim illustrates several advantages of this approach, with particular reference to co-growth bar graphs representing an efficient way to identify objects in an image. By continuously tracking the matching colors and having a fixed distance between them, the variable geometric information can be added to a regular color strip chart. Then, by considering color and geometric information, the object recognition method can still work under the condition that the background is confusing and the amount of occlusion and the object is buckling. However, the establishment of a model image database required for long-status chart matching is quite time consuming, and by calculating the model image and the search image, the distance of each possible unique non-arranged pixel is measured, and the co-growth bar graph is calculated. It also consumes considerable computational costs. In addition, the representation of the object does not include detailed geometric information other than the distance between the model and the searched image, and the information of the shape, size and position of the real object does not exist, so the final object position determination is not accurate, and subsequent The determination of the position continuously requires a large amount of calculation for 201040847. Furthermore, method parameters such as the size of the search window affect the object recognition accuracy, and the image must be resized so that the total size difference between the search and the model image is processed. One of the object recognition systems is effective and well-defined applications for the immediate identification of traffic signs on moving vehicles, and the traffic signal system must generally be able to quickly extract objects and properly classify objects. ❹ Ο In U.S. Patent No. 6,801,638, Jassen et al. describe a branch and device for identifying a traffic number tfe that can assist in displaying the slogan to a viewer by means of a memory. In the method and apparatus, the image is captured by an image sensor and analyzed and classified by a classifier in the information processing unit. Next, the composite image of the traffic number is generated, and the image is stored in the memory unit and displayed through the display unit. y The input image is first searched by color and/or spatial position information, and the area of average probability is used to determine the traffic number objects that may be included. The way in which the object is identified in the decision area is to classify the image areas by the hierarchical characteristics of the traffic signals in a hierarchical and sequential manner, such as to confirm the shape (circle or square) and symbols. Use the calibration procedure, etc. The classifier compares the input object characteristic data; the typical characteristic data set stored in the memory unit, and the object is recognized when the comparison distance is lower than the group threshold. "~, w Carrying rice training for several times, in order to make a difference in the image of the weather due to weather changes and light conditions. In addition, the point t is related to the shape of the person in the unit that stores the shape of the wheel "_". In general, the mark is off _

Correlation related classifiers may not be accurate or slow. Relevant discs, green J /, spurs, viewing environment and / or storage medium 8 201040847 The quality of the data. Improving the shape data in storage requires a lot of training or a large storage database. It: owed, large storage databases require more processing time for correlation processing. For example, circular and square objects appear in different (four) and rectangular shapes at different viewing angles, which may reduce the accuracy of object recognition. The above method also requires searching for the area of the traffic signal associated with the color value and/or spatial position that the input image may contain. In US Patent No. 6,813,545, stromme describes a system that alerts a driver to the presence of at least one particular traffic number, which consists of an image unit negative library, an automatic identification unit, a double selection mechanism, and a sound. And/or a visual indicator, wherein the image unit is connected to the road in front of the vehicle, the database includes at least one pre-stored traffic symbol shape, and the automatic identification unit is used to detect and confirm the continuous image. The traffic sign, and the detection and confirmation is achieved by searching for the shape in the database. The dual signal selection mechanism is included in the same image to determine the distance between the vehicle and the sign. 0 Sound and / or visual indicator notifications A confirmed traffic sign has appeared on the road in front of the vehicle. The input image is periodically captured based on the vehicle speed, and each input image is analyzed in a shape recognition processor to detect the shape of the traffic signal and the shape of the traffic symbol contained in the shape information. The shape search and recognition unit in the system uses a conventional image processing method such as Canny edge detection, followed by a simple pixel-by-pixel matching procedure on the processed image. The number of directions in the shape of the logo is stored in the shape matching database. For the triangle in the symbol, the 9 201040847 circular or rectangular symbol is also confirmed by the pattern or for the identification algorithm on the detected shape. The color _ was also executed to verify that the traffic number was true. However, edge detection and pixel-by-pixel matching methods are generally inaccurate and relatively slow. In addition, the difference between the shape of the same symbol and the position of the vehicle is quite large, which makes the shape matching and object recognition performed by edge prediction and pixel-by-pixel matching. Also, in U.S. Patent No. 5,926,564, MaSayuki Ki_ describes an image recognition in a long chart mode, however, it mainly scans for the X-axis and the Y-axis, and is "〇",,, i, , the method of image comparison 'such a comparison mode ... Once the image capture device ^ shot: When the object produces an angular difference, the shape and size of the object cannot be correctly identified. It can be seen that there are still many shortcomings in the above-mentioned methods of use, which is not a good design, but needs to be improved. In view of the shortcomings derived from the above-mentioned conventional methods, the inventor of the present invention has improved and innovated, and after years of painstaking research, he finally succeeded in researching and developing the method for identifying objects in the image. SUMMARY OF THE INVENTION The object of the present invention is to provide a minimum specific point and a maximum specific point obtained by the x-axis or γ-axis bar graph, and the polynomial regression analysis can quickly and easily determine the shape and size of the object and The method of identifying an object in an image. A second object of the present invention is to provide a method for recognizing an object in an image, which can capture different recognizable image features, such as RGB color, input image; step 2; for example, to obtain a digit

G Ο 201040847 Ge ash "like features or video image features, and quickly and accurately used for the identification of empty materials · ^ shape" can be applied to the father through the 5 ^, or other areas. Steps into the above purposes The method of recognizing an object in an image, which first scans the image data to capture the digital input image _ among them, an identifiable color shirt feature (such as RGB or IHS color) < sign or spectral feature, etc. ;,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, Step 4 Find the χ axis or γ axis strip point and the maximum specific point, in order to find the circle of the small special 疋 + m to find the U and other long strips of the material of the towel; point and ί large ^ Μ axis or 丫 axis length The smallest specific shape on the chart '· If (4) the object is linear, it can be judged as three objects = the straight line is judged as a square line; if, the right is a shape; the pay 0 - the human equation is judged as a circle 6 · Determining the size of the object by the distance between the two minimum special points on the x-axis or γ-axis strip chart; Take the J-point judgment = bit to borrow the smallest specific step-by-step 8-step image from the x-axis or γ-axis bar graph, and re-apply to identify other features in the image. 201040847 [Embodiment] The method (7) is not a process step diagram of the method for recognizing an object in an image according to the invention, and mainly includes: tr of the image, which can also be used to perform digital image data on the object. Paying for digital image data; in addition, the digital image is also taken by a class of image capture device; =r:r, the same can be taken - digital image == Ο 或 or analog image: capture digital image capture device image The device or camera or other admissibility, the operation of a single H RGB color is special - gray or, when the image or L7 can be said to recognize the grayscale features; The image data is the visible spectral frequency range, and the single-light m image can be _ feature; Step 3·· After the image identifiable feature is extracted, that is, 2: t take the X-axis direction (horizontal) or γ-axis of the image Direction (vertical) = graph:. 3′······························································································· After the axis obtains the X-axis or Μ long bar chart, the minimum specific point and the maximum specific point of the Χ-axis or Υ-axis bar graph are obtained, and the upper special point and the larger specific point are identified as identifiable features. The zero ^ = and maximum point pixel 8G4; in addition, the minimum and most search mode '« is found by the linear search mode on the Χ axis or = = 12 201040847' and simultaneously notes and records the minimum specific point and fixed point Position; in addition, can also be different search algorithm or method + = minimum or specific step in the chart data of the axis or the axis, and the maximum specific step l and then through multiple cut analysis methods, small specific points and maximum comparison Point; t derives the shape of the object 8G5; for example. If the minimum specific point to the maximum specific point is linear, then the object to be identified should be a triangle; ^ the minimum specific point to the maximum special = the line is a straight line Go to 'can determine that the object to be identified should be Square opening/or rectangle; if the minimum specific point to the maximum specific point is in accordance with the quadratic formula, it can be determined that the object to be identified should be circular or elliptical; *Step 6: According to the axis or γ of the image The position of the axis and the minimum specific point and the maximum specific point of the axis can be obtained as the pixel value in the direction of the 2 axis or the axis. The pixel size can be accurately calculated by the pixel value 806; ° Step 7 : According to the 彡 及 及 及 及 及 及 及 及 及 及 及 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定 判定Other characteristics of different colors or spectral frequencies), and the processing flow from step 2 to step 7 is performed. Thus, multiple identification functions are performed in the same image without multiple 5' identification features to achieve more accurate and faster identification. The purpose of the shape, size and location of the object. The following is a schematic diagram of the first embodiment of the present invention. The image recognition step is as follows: Step 1. As shown in FIG. 2, the white background has five kinds of digital image data. They are a red triangle 1, a red circle 13 201040847 2, a red square 3, a green triangle 4 and a blue square 5 · The image can be drawn by a digital computer, or through a digital image capture device Shooting; Flying 頚 ratio ~ Ο Ο Step 2: As shown in Figure 3, 'When digital image data is acquired, it can be decided to take the image with red features first according to the characteristics of the image. Therefore, there is red in Figure 2. The red triangle i, red circle 2 and red square 3 of the feature will be taken out and image recognition will be started instead of the red feature image, such as green triangle 4 and blue square 5 will be in the image capturing process. Removed from step; Step 3: Calculate the X-axis and Y-axis strip charts; as shown in Figure 4, simultaneously calculate the red (non-black) pixels on each line in the captured image in Figure 3 to obtain X Axis (horizontal) strip statistics , the X-axis strip chart will display the triangle strip chart corresponding to the red triangle 1, the red circle 2 and the red square 3, the curve of the long strip chart 2〇1 and the square strip Chart 3〇1, and the numerical values in the horizontal direction of the x-axis bar graph represent the number of rows, while the vertical direction represents the pixel value; the data displayed in the X-axis bar graph retains enough information to determine The shape of the object in the original input image and the size and position of the horizontal object; as shown in Figure 5, the red (non-black) pixels in each column of the image are captured in Figure 3 to obtain the gamma axis (vertical) Long bar chart; the γ-axis bar graph will display the triangular strip chart 102, the curve strip chart 202 and the square strip corresponding to the red triangle 丨, the red circle 2 and the red square 3 Chart 302, and the numerical values in the horizontal direction of the x-axis bar graph represent the number of columns, while the vertical direction represents the pixel value; the axis bar graph data holds sufficient information to determine the original input image. Object shape and vertical Object size and position 14 201040847 - Step 4. Then find the minimum specific point and the maximum specific point according to the x-axis or γ-axis bar graph. As shown in Figure 4, the two minimum specific points and one are obtained by linear search. The maximum specific point, at the same time will find the two small specific points and the position of the largest specific point (number of rows) annotation and record; take the long axis chart as an example, the pixel values of the two minimum specific points are all defects Therefore, the pixel search point is searched line by line toward the corresponding line number, and the first minimum specific point © A of the triangle strip chart 1〇1 can be searched in the 50th line, and the 50th line is noted. And the record, followed by the search for the second smallest specific point c of the chart is the 17th 〇 #, and = the specific point B is the pixel 85 _ point, and the specific points B, c of the search are noted and recorded; Then search for the curve bar 囷2〇1 and the square bar chart 3〇1 minimum specific point d'f'g] and the maximum specific point eh, I; wherein the minimum specific point D is 273 lines, 362 Line, g is 498 lines, J is 563 lines, and the maximum specific point E pixel is % point, η and I pixel is 86 points; After searching for the terminal of the axis long bar chart; the same as the triangle bar chart shown in Figure 5 shows the triangle strip chart 1G2, the curve bar chart 202 and the square bar chart The minimum specific point and the maximum specific point of 3〇2 are also obtained in the same way. Therefore, the minimum specific points κ and Μ of the triangular strip chart 1〇2 can be obtained, respectively, and 50 lines and 136 lines, respectively. The specific point L is the pixel 119 point; the minimum specific point p of the curve strip statistics о 202 is 160 lines and 245 lines, respectively, the highest specific point 〇 is the pixel 88 points; the minimum specific point of the square strip chart 302 (^ And τ are 28 lines and 365 lines respectively, and finally the specific points R and S are 65 points of pixels, and the terminals of the γ-axis long chart are sequentially searched; 15 201040847 Step 5 · After obtaining the minimum specific point and the maximum specific point , as shown in Figure 4, the regression analysis is performed through the X-axis bar graph to determine the shape of the possible object; it is recorded and recorded as the minimum specific point and the position of the largest specific point. Define the line 1 of the triangle strip chart 101 〇11 and 1012 and curve strip chart 2〇1 line 2011 and square strip chart 3〇1 line 3〇11,3〇12, 3013 boundary shape; first 'system for triangle strip chart 1 The results of the multi-parameter regression analysis of the line 1〇ii of 〇j:

Degree 1: -7〇.28 + 1.398x, a=〇.〇55 p < 〇.〇〇〇!, p<0.0001, R-2 = l.oo ;

Degree 2: -70.71 + 1.41x - 7.〇424e-5x^2, a=0.05, p<0.0001, PM.OOOl, p = 0.6085, ΙΓ2 = 1.00; ... the result shows that the two-dimensional strip chart may be Slope 〗 398 is linear with each other. The results of multiple regression analysis show that the second level (Degree 2) is statistically insignificant at a=0.〇5. The result of the polynomial regression analysis for the line 1〇12 is as follows: 〇 OQgTeQ U 237 2 - a = 〇〇5j p<〇.〇〇01, p<0.0001,ΪΤ2 = l.oo

Degree 2: 238.5 - 1.418X + 7.〇424e-5x^2, a=0.05, p<0.0001, p<〇.〇〇〇i, P=0.6085, R^2 = i 〇〇Results show triangular strips The statistical graph 101 may be lined with a slope of 1.398. The results of the polynomial regression analysis show that the second level term 2) is statistically insignificant at a=〇.〇5. Therefore, the combined result of lines 1011 and 1021 shows that the object in Figure 2 is a red triangle 1; 16 201040847 Next, the results of the polynomial regression analysis of the line 2011 of the curve strip chart 201 are as follows:

Degree 1: 48.71 + 0.05037x, a=0.05, p = 〇.0784, p = 0.5589 RA2 = 〇.〇〇 ;

Degree 2: -3331 + 21.48x - 0.03375x^2, a = 0.05, p<0.0001 p<0.0001, P<0.0001, ΙΓ2 = 0.95; the result shows that the result of the curve strip chart 201 is likely to be round or egg The shape-polynomial regression analysis showed that the second-order term (Degree2) 在 was statistically significant at a=0.05. Therefore, the result shows that the object in Figure 2 is a red circle 2. Then, since the lines 3011, 3012, and 3013 of the square strip chart 301 are vertical and horizontal lines, the polynomial regression analysis cannot be performed; however, the line 3011 represents a line from the minimum specific value of the zero point to the maximum specific value, and the line 3012 contains all equal maximum specific values, and line 3013 represents a straight line from the maximum specific value to the minimum specific value of zero value, so the constituent objects of the lines 3011, 3012, and 3013 are a red square 3. 〇Step 6: After knowing the shape of the object, determine the size of the object, as shown in Figure 4. The minimum specific points A and C of the triangular strip chart 1〇1 are 5〇 and n〇, the maximum The specific point B is 85 pixels, so the actual size of the red triangle 1 can be determined by the camera and image characteristics and the position of the camera in the actual environment. 〃 , the minimum specific point D, F of the curve bar graph 201 is 273 ^ and 362 lines, and the maximum specific point E is 86 pixels. Therefore, if the size of the red circle 2 of Figure 1 is to be calculated, the camera and Image characteristics The position of the camera in the actual environment is determined. 17 201040847 The minimum specific points G and J of the square strip chart 301 are 498 lines and 563 lines, and the maximum specific point H and the work are 86 pixels. Therefore, if the size of the red square 3 in Fig. 2 is to be calculated, It can be determined by the camera and image characteristics and the position of the camera in the actual environment. Step 7. Finally, calculate the position of the object. Please refer to Figure 4 and Figure 5 at the same time. The minimum specific points a, c, d, f, g obtained through the X-axis strip chart and the ¥-axis strip chart. , j, k, m, n, p, Ο Ο Q, t determine the position of the object, wherein the X-axis minimum specific point A of the triangular strip chart 101, (^ is 50 to 17 〇; and the Y-axis The minimum specific point K and Μ are 52 1 136'. The minimum specific points A, C, K, and X of the X-axis and the γ-axis can be used to know the position of the triangle strip chart. The curve strip chart 2 (1) X-axis minimum specific point D, U 273 to 362; and Y-axis minimum specific point N, P is 16〇 to 245, through the X-axis and Y-axis minimum specific point D, F, N, p data , you can know the position of the curve strip chart; the square strip statistics ® 3〇kx axis minimum specific point (1) is Γ to 563; H axis minimum specific point Q, T is 365, through X axis and Y The most axis, j, shoes & ρ ρ wrong "... J, Q, T data, you can know the position of the square strip chart; and ~ two if necessary, the size of the real object can usually be seen by the camera and shirt Characteristics and photos The position of the camera in the actual environment is determined. Step 8: The green identification image data can be exchanged, and the shape, size and position of the identified object of the color recognition feature can be repeated. The image recognition of this month is completed under the condition of no pattern comparison and pattern ratio, and the object identification method improves the speed and accuracy of object utilization. In addition, the object shape information is requested. Out, the W 2 regression analysis is performed, so the RA2 is suitable for analysis to provide the measurement of accuracy: In addition, there is no accurate and time-consuming direct input image search when identifying the shape of the object. 6 to FIG. 9 is a second embodiment of the present invention, which is mainly applied to traffic image recognition, and the main identification step is ··m

Step 1: As shown in Fig. 6, the traffic sign is taken by the image capturing device to obtain the traffic sign image 6 of the road. The image of the traffic signal 6 has a red outer ring 61 and a white inner ring 62. And black font 63 and other image-recognizable features' At the same time, the traffic number images 6 also contain other red features, such as: safflower 71 on the tree and a plurality of safflowers in the shrub 7 safflower 71 some parts are green leaves Obscured; ... Step 2: As shown in Figure 7, the image data with red features in Figure 6 is captured, so the red outer ring of the traffic sign image 6 and the red flower 71 on the tree and the safflower in the bush can be obtained. 71 and other image features; Step 3: As shown in Figure 8 and Figure 9, the X-axis and the Υ-axis strip graph are obtained by calculation. The χ-axis and γ-axis strip graphs contain red outer loop strip statistics. Figure 501, safflower statistics on the tree 5〇2 and a plurality of safflower statistics 503 on the bush, wherein the unique figure of the long chart 5〇1 is a feature of the shape of the original annular object; The axial strip chart shows that the red outer ring 61 and the safflower 71 on the tree are at γ = In the (vertical) direction, the red outer ring strip graph and the red flower strip graph 5〇2 data are added to each other to be in the red outer loop strip graph 501 corresponding to the object to be identified. Adding the noise to another 19 201040847, the unique figure of the long chart 501 is the feature of the shape of the original ring object; the calculation mode of the above X-axis and γ-axis bar graph is the same as that of Figure 4 and Figure 5. As shown in Figure 8 and Figure 9, the X-axis and Y-axis long chart data are scanned from zero line to line to find the minimum specific point of the 红色 axis red outer ring strip chart 501. a', D, and maximum specific point Β, C' and Υ axis red outer ring strip chart 〇1 of the minimum specific point Ε, Η, and the maximum specific point F,, G,; the minimum specific point Eight, 1U2 lines, D, 〇 is 2040 lines, Ε ' is 1099 lines, Η, is 2078 lines; the maximum specific point Β 'and C' pixel values are 710 points, and F, and G, the pixels The values are 66〇 and 650; in addition, the minimum specific points a, D, Ε, Η, and the maximum specific points B', C', F, G, The search methods are the same as those in Figure 4 and Figure 5, and are not described here. Step 5: After obtaining the minimum specific point and the maximum specific point, use the three sets of lines shown in Figure 501. 5012, 5013 with polynomial regression analysis (regression) Judging the shape of the object; the polynomial regression analysis of line 5011 in Figure 8 is as follows: 〇Degree 1: -4439 + 4.145x, a=〇.〇5, p < 〇〇〇〇lj p<0.0001, RA2 = 0.93 5

Degree 2: -42765 + 68.91x - 0.02733x^2, a=0.05 p<0.0001,p<0.0001,p=0.0001,ΙΓ2 = 0.99 The result shows that the line 5011 is likely to be circular or _-shaped and polynomial regression analysis results It is shown that the 'second level term is statistically significant at a = 0.05; the polynomial regression analysis of line 5012 in Figure 8 is as follows:

Degree 1: 352.3 + 0.008745x, a = 〇.〇5,p<〇 〇〇〇1 20 201040847 ρ = 0_6111,RA2 = 0.00

Degree 2:6456 - 7.86U + 0.〇〇25〇3χΛ2, a=〇05, ρ<0.0001, ρ<〇.〇〇〇1,ρ<〇.〇〇〇1,RA2 = 〇89 ' Line 5012 may be round or oval, and multiple under regression analysis results show that 'the second level term is statistically significant at a=0.05; the line 5013 polynomial regression analysis in Figure 8 is as follows:

Degree 1: 9155 - 4.43x, a = 0.05, p<〇.〇〇01>p<〇〇〇〇1> RA2 = 0.92 ’ Ο

Degree 2: -102241 +109x - 0.02887x^2, a=〇.〇5, p<0.0001, p<0.0001, p=0.000l, R-2 = 0.99 ' The result shows that the line 5013 may be round or oval And polynomial regression analysis, the second-level term is statistically significant at a=〇.〇5; > Therefore, the combined result of the line 5〇1 Bu 5〇12, 5〇13 shows that Figure 8 shows The long bar graph 501 is likely to be a circular or oval shape of the annular body. Step 6: After knowing the shape of the object, determine the size of the object, as shown in Figure 8, the red outer ring in the X-axis direction. The minimum specific points A' and D of the long bar graph 5〇1 are 1112 lines and 2040 lines, and the maximum specific points B' and C' are 710 pixels; and the minimum specific point E' of the x-axis direction shown in FIG. H' is 1〇99 lines and 2078 lines, the maximum specific point F is 66〇 pixels, G is 650 pixels, and the correct size of the object can be calculated through the above conditions; in addition, the correct size of the object It can be determined by the camera and image characteristics and the position of the camera in the actual environment, and will not be described here. /Step 7. After the dare, the position of the nose object, please also refer to Figure 8 and Figure 9, through the X-axis strip chart and the Y-axis strip chart 21 201040847 The minimum specific point a, d,, e, h, ^ u , the judgment of the position of the entering object, the ton w breaking mode is not repeated with the above figure 4 and Figure 5; (4) The way of the road is the same 'this and repeating the shape, and may be multi-step Step 8: Re-take the other image identifiable features Step 2 7 to re-identify the image. Therefore, information such as the size and/or position of the object obtained by the method of the present invention can be used for other uses or fields in the traffic number Wei. Ο 〇 In addition, the safflower chart on the tree shown in Figure 8 is in the 〇2"" ^ ^ analysis shows that the object does not have a well-defined linear or curved line because the ΙΤ2 value of the first- and second-level analysis is small:

Degree 1: _1653 + Q 5543x, a=()()5 〇 p<0.0001, R^2 = 0.48

Degree 2: -155477 + l〇2x - 0.01672x^2, a=0.05, P<0.0001, p<〇.〇〇〇is p=〇>0〇olj R^2 = 0.66 Therefore, the object is obviously not A circular or linear traffic number object. When using the polynomial regression to analyze the image shape, it is not affected by noise. As shown in Figure 9, the polynomial regression analysis results for the line 5〇14 displayed on the γ-axis strip chart are as follows:

Degree 1: 393 _ 〇.〇2945x, a=0.05, p<0.0001, p=0.0370, R^2 = 0.01

Degree 2: 5350 - 6.353x + 〇.〇〇1987xa2,a=0.〇5, p<0.0001,p<〇.〇〇〇l,p<〇〇〇〇1,RA2 = 〇9 result shows strip The line 5014 of the statistical graph 501 may be circular or oval (not linear)' and the result of the polynomial regression analysis shows that the second level term is statistically significant at a=0.05, which is the same as the graph of the X-axis strip chart of FIG. 22 201040847 2 5012 is similar, so the shape identification of the long axis chart is not affected by the noise of the red flower on the tree. & This (4) Correction has the following advantages when comparing the techniques of the f彡icon identification (4) materials: /, / The most special points and m points obtained on the small extra large axis or γ axis long bar chart 'With polynomial regression analysis to determine the shape, size and position of the object in the image

Ο 2. The invention can capture different identifiable image features

... to make a payment, such as RGB color feature money image features or video image features, and to identify the shape, size and position of the captured image, so that it can be applied to traffic numbers or other fields. The detailed description of the above is a detailed description of the possible embodiments of the present invention, which is not intended to limit the scope of the invention, and the equivalents or modifications thereof The patent scope of this case. It seems that the above-mentioned materials (4) are technically (4), and can improve the above-mentioned multiple functions compared with the conventional articles. 'The statutory invention patents that should fully comply with the novelty and the progressiveness, and apply in accordance with the law, please ask your office. Approving the invention patent application for this invention, in order to invent the invention, to the sensation of the child, the moon [a simple description] " Figure 1 is a step diagram of the method for identifying an object in the image; Figure 2 is the identification of the image in the present invention The color of the five objects in the first implementation of the method of the object; FIG. 3 is a schematic diagram of the identifiable image features in the second drawing; FIG. 4 is the X-axis color length obtained from the color pixels in each row of the figure III. Figure γ-axis color length Figure 5 is a bar chart obtained by calculating each column of color pixels in Figure 3. Figure 6 is a second embodiment of the color traffic number of the method for identifying objects in the image; Figure 7 is a screenshot Figure 6 is a schematic diagram of the identifiable image features; Figure 8 is a statistical diagram of the χ-axis color strip obtained by calculating each color pixel in Figure 6;

Ο Figure 9 is a γ-axis color bar graph obtained by calculating the color pixels of each column in Figure 6. [Main component symbol description] 1 red triangle 2 red circle 3 red square 4 green triangle 5 blue square 6 traffic standard 61 red outer ring 62 white inner ring 63 black font 71 safflower 101 triangle strip chart 1011 line 1012 line 102 triangle strip chart 201 curve strip chart 24 201040847 2011 line 202 curve strip chart 3 01 square strip chart 3011 line 3012 line 3013 line 3 02 square strip chart 501 outer ring strip statistics Figure 0 5011 line 5012 line 5013 line 5014 line 502 tree safflower strip chart 503 shrub safflower strip chart A, C, D, F, G, J minimum specific point B, E, H, I maximum specific point K, Μ, N, P, Q, T minimum specific point QL, 0, R, S maximum specific point A, D, E, H, minimum specific point B, C, F, G, maximum specific point 25

Claims (1)

201040847 VII. Patent application scope: 1. A method for recognizing an object in an image. The step η is first obtained - the digital image data is required to be d ^ 2. The digital image can be recognized by the function; :::: simultaneously calculated Step 2: Take the image of the long axis of the image; after obtaining the X-axis or γ-length bar graph in red 4', find the X-axis Ο Ο: the minimum specific point and maximum of the γ-axis long bar chart The specific point; Wonder 5, n through the polynomial regression analysis method, according to the characteristics of the identifiable object to determine the shape of the object. 2. If the application of the patent specification (4) is described in the category of the towel identification object, it further includes a step 6, which is to capture the image data of the same image but different identifiable features in step ι, and repeat the steps. Process flow from 2 to 5. 3. The method for recognizing an object in an image as described in the first aspect of the patent application, wherein the digital image data obtained in the step 1 is obtained by photographing the object by a digital image capturing device. 4. A method of recognizing an object in an image as described in claim 3, wherein the digital image capturing device is a camera or a camera. 5. The method for recognizing an object in an image according to the first aspect of the patent application, wherein the digital image data obtained in the step 1 is obtained by capturing an object by an analog image capturing device, and converting the analog signal into Digital signal to obtain a digital image. 6. The method of recognizing an object in an image as described in claim 5, wherein the analog image capturing device can be a camera or a camera or other device capable of capturing images. 26 201040847 7. The range of objects identified in the image as described in item 1 of the η 利 范围 肖 肖 步骤 Step 2 The digital image captured in step 2 can be identified as a single color RGB or IHS color identifiable feature. The method of recognizing an object in the image as described in the first item of the Fashen range is identifiable. 2 The digital image is identifiable. The identifiable feature is ash 9.m. The range is identified in the image. Ο Ο 频 四 (4) 步骤 Step 2 (4) Digital image identifiable features are identifiable features of the video pre-spectrum frequency. 10 ϋ 帛 帛 在 在 在 在 在 在 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识Pixel. 11. If the scope of the patent application is specified, the minimum specific point of the object identified in the image or the axis i of the axis = the fourth step is obtained in the linear search mode. The obtained zero Ο 12·Γ=range The first specific point of the step 4 of recognizing the object in the image as described in the first item is obtained by linear search on the X-axis 点 pixel, / on the graph The maximum value obtained in the model 13: the square position of the object identified in the image as stated in the application: the minimum specific point and the maximum specific point of the pure note and 2 are as follows: Where the step 5 = = the equation of the object is determined by the regression method, and the line on the X-axis or Y-axis strip chart is analyzed according to the system 27 201040847, from the minimum specific point to the maximum specific point The system is linear, and it can be determined that the object to be identified should be a triangle. • A method for recognizing an object in an image as described in claim 13 wherein the line from the smallest specific point to the maximum specific point is linear, and the object to be identified should be square or rectangular. It can be determined that the object to be recognized should be a circle or an ellipse in the method of recognizing an object in the image described in the thirteenth application patent, wherein the minimum specific point of the line to the maximum specific point is a quadratic equation. shape. 17. The method for recognizing an object in the scene includes: Step 1: acquiring a digital image first; Step 2. extracting the identifiable feature of the digital image; Step 3·· simultaneously calculating Step 2: Take the x-axis or γ-axis long circle of the image; Step 4: Obtain the X-axis or γ-axis bar graph, and then find the minimum specific point and maximum of the x-axis or x-axis bar graph. Specific point; ❹ Step 5. Then through the polynomial regression analysis method, determine the shape of the object according to the characteristics of the identifiable object; Step 6: Determine the size and size of the object according to the minimum specific point and the position of the largest specific point. 18. The method for recognizing an object in an image as described in claim 17 of the patent application, further comprising a step 7 of extracting image data of the same image but different identifiable features in step i, and Repeat the process flow from step 2 to step 6. 19. The method of claim 2010, wherein the digital image obtained by the step 1 is obtained by photographing the object by a digital image capturing device. 〇.=申凊专(四), in the method of identifying objects in the image as described in Item 19 of the 19th, the digital image capturing device is a camera or a camera. 21. = Shen Qing Patent Range No. 17 The Ο Ο method for recognizing an object in the image is obtained by capturing the object image by the class-like operation device, and converting the analog signal into a digital signal to obtain a digital image data. 22. The object of recognizing an object in an image as described in claim 21 of the claim is wherein the analog image capturing device can be a camera or a camera or other device capable of capturing images. A method for recognizing an object in an image as described in claim 17 wherein the digital image captured in step 2 is identifiable as a single color RGB or IHS color identifiable feature. The method of claiming the patent image is as follows: in the image method, the image recognition object of the image is the grayscale identifiable feature. 25. The method of identifying an object in an image as described in claim 17, wherein the digital image recognizable feature of the step 2 is a video spectral frequency identifiable feature. The method of recognizing an object in an image as described in claim 17 of the patent scope, wherein the calculation of the X-axis or γ-axis bar graph of the (4) 3 is to calculate a pixel on each line of the image recognizable feature. /A The acquisition of the minimum specific point of the 2010 201084847 _ Λ 4 of the object identified in the image as described in claim 17 is obtained in the linear search mode on the zero-pixel stripe statistical circle of χ 28·=Please take the maximum specific point of the 4th step 4 of the station red in the image as described in item 17 of the scope, which is obtained in the linear search mode on the X-two pure bar chart.豉 Large value points pixels. The method of recognizing an object in an image as described in Item 17 of the patent scope, the smallest specific point and the maximum specific point position obtained by the towel (4) 4 must be noted and recorded. 3〇.=Please refer to the paradigm in the patent scope to identify the object in the image. The polynomial regression analysis method of step 5 is based on 2 to analyze the lines on the X-axis or γ-axis strip chart, when the line From the minimum specific point to the maximum specific point, the linear trend is obtained, and it can be determined that the object to be recognized should be a triangle. 31. The method for recognizing an object in an image as described in claim 29, wherein the minimum specific point to the maximum specific point of the line is a straight line, and the object to be identified should be square or rectangular. It is determined that the object to be identified should be circular or by the method of identifying the object in the image as described in item 29 of the patent scope, wherein the minimum specific point to the maximum specific point of the line = the sub-equation Oval. 33·~ The method of identifying objects in the image mainly includes: V1 is to obtain a digital image first; Step 2: to extract the identifiable features of the digital image; 30 201040847 Step 3: Calculate the steps at the same time 2 The axis or γ-axis bar graph of the captured image; Step 4: After obtaining the X-axis or Υ-axis bar graph, the minimum specific point and maximum specificity of the χ-axis or Υ-axis bar graph are obtained. Point 5: Determine the shape of the object according to the characteristics of the identifiable object through the polynomial regression analysis method. Step 6: Determine the size of the object according to the minimum specific point and the position of the largest specific point; Ο 34 35. Ο 36 37. Step 7. Determine the correct position of the object based on the position of the smallest specific point and the maximum specific point. The method for identifying an object in the towel as described in Item 33 of the application (4) further includes a step 8 of re-acquiring the image data of the same image but different identifiable features in step i, and repeating Process flow from 2 to 7. ^ The number of image recognition objects in the image as described in item 33 of the patent application scope, and the digital image data obtained in the step 1 are obtained by photographing the object by the digital f image capturing device. A machine for identifying an object in an image as described in item 35 of the benefit range. ', the digital image manipulation device is the camera or the photography method (4), the ratio of identifying the object in the image as described in item 33, the digital image f obtained in step H is replaced by the class The object is captured and the analog signal is converted to obtain the digital image data. The analog image capturing device for recognizing an object in an image as described in item 37 of the scope can be a camera or a camera 31 38. 201040847 or other device capable of capturing images. 39. The method of identifying an object in an image as described in claim 33, wherein the digital image captured in step 2 is identifiable as a single color RGB or IHS color identifiable feature. 40. The method for recognizing an object in an image as described in claim 33, wherein the digital image recognizable feature captured in the step 2 is a gray-scale identifiable feature. 41. The method for identifying an object in an image as described in claim 33, wherein the digital image identifiable feature captured in the step 2 is a video spectral frequency identifiable feature. 42. The method for recognizing an object in an image as described in claim 33, wherein the step of calculating the X-axis or γ-axis strip chart of the step 3 is to calculate each line of the image recognizable feature. Or pixels on each column. The maximum point pixel. 45. If the patent application scope % method 'where the position of the step 4 must be noted and recorded 46. The minimum specific point and maximum specificity of identifying the object in the image as described in claim 33, paragraph 33 The 32 201040847 method for recognizing an object in an image, wherein the polynomial regression of the step 5 is a large a ^ v 4 ah, according to the line on the X-sleeve or Y-axis strip chart. Eight analysis ^ The minimum specific point of the line to the maximum specific point is linear, ^ can be determined that the object to be identified should be a triangle. 47. The method for recognizing an object in an image according to claim 45, wherein a line from a minimum specific point to a maximum specific point of the line is determined to be a square or a rectangle. . 48. The method for recognizing an object in an image as described in claim 45, wherein the minimum specific point to the maximum specific point of the line conforms to -Λ>-fc- The object to be identified should be circular or elliptical. 33