TW594591B - Image-based protruded part recognizing and tracking method - Google Patents

Abstract

There is provided an image-based protruded part recognizing and tracking method, which comprises: converting an image for being expressed by HSV image format; recognizing the characteristic region in the image based on a hue range and capturing the regional profiles corresponding to the characteristic region; for every pixel of the regional profile, determining whether the distance between the pixel and the gravity center of the characteristic region is larger than the distance from the previous one of the pixel and the next one of the pixel to the gravity center. If the distance between the pixel and the gravity center of the characteristic region is larger than the distance from the previous one of the pixel and the next one of the pixel to the gravity center, it is determined that this pixel is a protruded part in the characteristic region.

Description

594591 V. Description of the invention (i) The present invention relates to an image recognition and tracking method, and an effective identification and tracking of the prominent parts in a particular image based on the hue of the image and based on the highlights of the protruding parts. Identify and track the two shadows. It is based on the ~ image. For application developers, in addition to application considerations, the interaction between the program and the user, the strength of the factors. The key interface between the general program and the user is the key test and mouse. The user can use keyboard and mouse to give two or more keyboards for related operations and calculations. In addition to the 7 application process, with continuous innovation and research and development; a new generation of human-machine interface based on human beings has also been tried to develop. ^ Immature, this new generation based on " hand ": Machine =: Due to the special technical environment, such as 'must be maintained on the reflection plane ^ = limited to interference in a very small environment, but' such restrictions will reduce the use of production to reduce the use of hands •, based on The interface must be created in advance for subsequent segmentation, which increases its inconvenience. In addition, if skeleton is used as the feature of gesture recognition, it is easy to burn to the influence of noise 0 ^ 6 > and if contour is used to identify the car, it is more difficult to complete the contours of hands and body parts. segmentation. In view of this, the main object of the present invention is to provide a method for identifying and tracking parts based on image (Hue) based on the protrusions in the two-tone tone in the characteristic area. % ≪ 大 出 In order to achieve the above-mentioned object of the present invention ', the present invention can provide

594591 V. Description of the invention (2) Image-based identification and tracking of prominent parts According to the embodiment of the present invention, the image-based method is used. -. Tracking method. First, an image is converted to a point ▲ ° 卩 position identification and (HSV) image format for representation, according to a color tone, saturation, and brightness feature area, and the corresponding feature area is captured. Identify each pixel point on the outline of the area in the image, and determine whether the distance after the symplectic point ^ is greater than the distance between one pixel point before the pixel point and 1. If the distance from the pixel to the center of gravity is greater than the distance from the image center to the center of the next pixel to the center of gravity, then the rule is the protruding part of the pixel. j 疋 This pixel point is a characteristic area. In addition, the present invention receives another image, and recognizes it by v 从而, so as to identify the protruding part and the center of gravity of the protruding large image in this image. If the two shadows are less than a predetermined angle threshold and two shadows; when the difference between the missing angles is greater than a predetermined distance threshold, /, a distance of ~ is triggered by a specific application or function instruction. A trigger message, taking the embodiment x, and the first figure as a flow chart, showing the implementation of you in accordance with the present invention. Even the identification and follow of salient parts based on the horse ceremony: Yes, this embodiment is based on the hand Process on top of the area. Note the identification and follow-up of the protruding part, but not "in this part of the dog" as an example. First, if step S10 is involved, the first image is received and converted into hue and saturation Λ /; 2; step size 1, It will be shown. Note that if the image format (HSV) image format is "image received, it is hue, saturation, and brightness. Page 5 0213-8797TW (nl); STLC ~ 01-D ^ 9118; YIANH0U .ptd 594591

When the image format is indicated, step S11 can be omitted.着 ^ 'Identify the characteristic area in the image according to a tone range according to step s 12' and extract the corresponding feature area's area contour 1 according to step S1 3 because this embodiment is used to identify the hand area. Fingertips, so the feature area is the hand area, and the tone range is the skin tone range. Suppose PixeU i, j) represents the hue information of the pixel in the image at the coordinates (i,] ·). Then use the predefined maximum value Hue-Max and minimum value Hue-Min (tone range) to cut features. Region, as follows:

Pixel {ij) ^ iFo ^ Sr〇und if Pixel {ij) < Hue ^ Max & Pixel {ij) ^ Mm l Background otherwise When the hue information of a pixel falls within the hue range, the pixel in this pixel sign region; Otherwise, it does not belong. 'After the temple, as in step S14, for the pixels on the outline of the area, determine whether the distance from the pixel to the center of gravity of the feature area is greater than before the pixel—the distance from the prime point to the center of gravity of the next pixel. If the distance from the pixel to the center of gravity is not greater than the distance between the pixels; ^ one pixel or the distance from the next pixel to the center of gravity (No in step S14), the determination in step S16 is performed (explained later). The distance from the 2 pixel point to the center of gravity is greater than the distance from the previous pixel point to the center of gravity—the distance from the prime point to the center of gravity (YES in step S14), then in step S15, it is determined that the 2 pixel point is a protruding part in the feature area, that is, the fingertip. . Fig. 2 is a schematic diagram for explaining detection of a protruding part. In this example, PiXel_CenterC (P_CC) represents the center of gravity of the feature area;

594591 V. Description of the invention (4)-1 乂 61-(: 0111: 0111- (〇 (? 一 (: (〇) represents the first contour point (pixel point) on the area outline; Distance (i) = (? 1 and 6 1 1 (: 0111: 0111 '(1)-? 1 乂 6 1 一 〇61116 1 ^) 2,1) (1) represents the distance from the first contour point to the center of gravity. Determine the contour point The formula for whether it is a relative extreme value (highlighted niches) is as follows:

Pixel One Coniour, i) = ExiremaPomt; if Distmce {i ^ \) < Distance {i) 8l Dis tm ce [i ^ \) < Dis ce {i) where ExtremePoint represents the feature point (protruding part). When the distance from the Luo point to the center of gravity is greater than the distance from the neighboring contour point to the center of gravity, the 1 contour point is determined as a feature point. < Then, as in step S16, it is judged whether or not the judging portion is judged on the outline of all the regions. If not all the contours of the area outline are used to judge the protruding part (No in step S16), then go to next pixel point above and below the area outline, and go back to the step to continue judging the pixels to the feature area. Before the center of gravity of the pixel point-Is the distance between the pixel point and the rear point greater than the silly sound on the outline of the area? Go to 1 = 0 < the distance from the pixel point to the center of gravity. And if the image is taken, when tracking the out part, when receiving another feature area: you can use the image tone range (skin tone range) to identify 2 as the range. The characteristic area in this shirt image, which is 〇213-8797TW (ni)

Ptd page 7 V. Description of the invention (5) 疋 The area scope is the area of the corresponding package A and this area. ::, all of the mid-feature areas-protrusions Identify the image with a predetermined area range 2: ::: set. ί Idea: The purpose of the 22 whole image is to spend a lot of time to tamper with it. The purpose of specializing the Q field is to avoid searching the whole image of the brother. When you find the special: J: the original image must scan the predetermined area to identify the feature = After that, other images can display the feature area range 30 including all feature areas 31 detected once and the feature points a, b, c, d ^ according to Figure 3a; The area range is 30 domains, with high H and wide W. Because the person affiliates the main A, e, and f & small, and there is a range of a certain range between the two consecutive images, a large block range for searching, such as :: mine definition-: near the existing location shift d area Position, the range of singularity / domain range 30 is -um) from the surrounding neighborhood. ., The position 'that is, the tracing area 32 has a height (H + 2d) and a wide value. Note that when the first shot converts the entire image to a color n-shirt image, it can be indicated, or only For Shadow 2 :: Degree ^ Degree (bile) image format, saturation, the meaning of the existing area range is converted to hue saturation, degree-free scene mu. * For the trigger mechanism on the surface, identify at the feature point to determine whether the change in the production position is used as a feature '. Figure 4 is a schematic diagram to illustrate the identification of the trigger mechanism. If

0213-8797OV (nl); STIjC.〇l.D.9118; YIANHOU.ptd page 8 594591 V. Description of the invention (6), and Theta (i) represents the angle between the feature point and the direction of the center of gravity connection. If the position change of the feature point satisfies the following conditions: ^ Green and green trigger mechanism (similar to the action of pressing the left mouse button). Is now Kai

Trigger = true, if (abs {7heta {iyneta {i ^)) < A ^ leThreshold) & iMMag {iyUag {i ^)) > MagUreshold) where AngleThreashold is the predetermined angle threshold,

MagThr easho Id is the established threshold. In other words, the direction of the connection between the feature point (protruding part) and the center of gravity in the first degree is connected to the horizontal right second image angle and the feature point in the first image is connected to the center of gravity ^ | ^ The difference between the included angle is less than the given angle threshold Value and the second image says 'the distance between the center of gravity and the distance between the protruding part and the center of gravity in the first image and-: a fixed distance threshold' means that the trigger mechanism is activated, which is triggered when a specific application or function command is triggered . Triggering It is important to note that the design or modification of the application of the trigger mechanism is not limited to this: In addition, due to the judgment described in the feature area Lu Xi 歹, the feature points can be filtered again ^-feature points due to trigger content and trigger conditions. Therefore, by using the identification and tracking method provided by the present invention, it is possible to use the characteristics of the protruding parts in the protruding part feature area based on the shadow two as the basis and to identify them based on the special identification and track. M ^ M + is effective

594591

In addition, the present invention has the following advantages: first, the use of color riding information to score ^ i can reduce the degree of influence by light, and because the cut is directly set in the feature area to be searched (in this embodiment, the hand) In the hue region, even in a non-specific environment, the feature region can be directly segmented without going through the steps of establishing a background type. Second, in this embodiment, the feature points with = fingertips in addition to having clearly identifiable In addition to the advantages, the processing in the positioning and tracking phase is simpler and saves system resources than the contour or backbone based approach. Although the present invention has been disclosed in the preferred embodiment as above, it is not intended to limit the present invention. Anyone skilled in the art can make some changes and turbid decorations without departing from the spirit and scope of the present invention. -The scope of protection of the invention shall be determined by the scope of the attached patent application. Cut into the mold side for precise protection

0213-8797TW (nl); STLC-01-D-9118; YIANHOO.ptd Page 10 594591 The diagram is briefly explained in order to make the above-mentioned objects, features and advantages of the present invention more clear and understandable. The example and the accompanying drawings are described in detail as follows. Figure 1 is a first-class red picture showing the operation flow of a method for identifying and tracking salient parts based on an image according to an embodiment of the present invention. Fig. 2 is a schematic diagram for explaining detection of a protruding part. Figure 3a shows the characteristic region range in the first image. Figure 3b shows the image area range of the detection feature area in the second image. Figure 4 is a schematic diagram for explaining the identification of the moon trigger mechanism. S1 0, SI 1,…, S1 7 ~ operation steps; two-a, b ....... feature points (protruding parts); 3 0 ~ feature area range; '31 ~ feature area; 3 2 ~ scan , Tracing area.

IH 0213-8797TW (nl); STLC-〇l-D-9118; YIANH0U.ptd $ 11 pages

Claims (1)

594591 Enclose the following column :: Recognition and tracking method of salient parts based on the image-the method package receives a first image; converts the first image to be expressed in the format of hue and ^ image; X, 凴 degree (1 ^ \ 〇 shadow domain identification-identifying a characteristic region in the first image from a tonal range to capture one of the first characteristic regions first, for each I region contour on the first region contour; and determine the pixel The point i should be greater than the distance from the center of gravity of the previous pixel to the pixel. The distance from the center of gravity of the pixel is from the distance from the pixel to the center of gravity. If the distance from the pixel to the center of gravity is greater than the prime and the next pixel The distance from the center of gravity is the protruding part in the feature area that is like a pixel before the two pixel points. Set the pixel point to be the second one. The method of identifying and tracking the position as described in item 丨 of the scope of patent application It also includes the following steps: image-based highlighting and the following steps: · Receive a second image, "Set a predetermined region in the second image to a second characteristic region in the second image, and its $ M The tone range identification Corresponds to the area containing the first pull M = in the first image, the predetermined area range and the position of a predetermined position around the area, and the corresponding one of the second feature areas is extracted. A one-by-one area contour on the second area contour; and ~ one pixel point, The scope of the six patent applications determines whether the distance from the pixel to the second feature area is greater than the pixel. The distance between the center of gravity and the center of gravity is distant; and the distance between the pixel-point and the back-pixel-point to the center of gravity is The heavy distance is greater than the pixel point and the next pixel $ 2: Before the two pixel points-the protruding parts in the characteristic region of the second image. "The pixel point is the image-based protruding rc method described in item 2 of the patent scope. It further includes converting the second image into a color saturation, shell degree (HSV) The image format is used to represent it. 4 · The image-based ::: identification and tracking method described in item 2 of Shenjing's patent scope, which includes converting the second image to the given area and expanding it to hue, saturation The image format is expressed in degrees and degrees (HSV). The distance between the position and the center of gravity has a predetermined distance threshold from the first shadow. The image-based highlight is described in 5. As described in item 2 of the scope of patent application The image-based identification and tracking method of protruding parts further includes if the angle between the protruding part and the center of gravity connection direction and the horizontal line in the second image and the protruding part and the center of gravity connection direction and the horizontal line in the first image The difference between the cool angles is less than a predetermined angle threshold and the difference between the distance between the protruding part in the second image and the center of gravity in the second image is a trigger message. 6 · As in the first item of the scope of patent application π winter hook Bit identification and tracking method, in which the tone range is a skin tone tone range7. The image-based highlighting 0213-8797TW (nl); STLC-01-D-9118; YIANHOU. ptd The 13th tribute application patent scope 8 discrimination: ί =: t, where the -feature area is the hand_area. Part identification and tracking Fang Wai: 1 item Institute: of: 乂 image-based highlight 9 9 w ', Where s mysterious dog's out part is the fingertip part. Including the following steps, such as the method of identifying and tracking salient parts based on the image, including the tone-saturation, saturation, and brightness (HSV) image format representation:-recognize the first The hand area in the image; the hand outline of the hand area; and for each pixel point on the hand outline, determine the pixel point to the hand area one pixel before the pixel point:; = away Is the distance from the Beijing point to the center of gravity to the right; the distance from the pixel to the center of gravity by I is greater than the protruding portion of the pixel S prime point and the next pixel to the center of the image / pixel point. Away ', the pixel is determined to be the hand ίο. Please refer to the 9th patent area identification and tracking method of patent scope. The method includes the following steps: receiving the second image based on hue, saturation and brightness; "Region area ::: area" where the range of the given area: the position to move; Dagger Grave 594591 Patent application scope and extracting the hand contour corresponding to the hand region. For each pixel point on the hand contour, judge the pixel point to the hand region, one pixel point before the pixel point and the latter image. : Whether the distance is large and the distance from 4 to the center of gravity; if the distance from the pixel to the center of gravity is greater than the distance from the pixel to the pixel + a prime point and the next pixel to the center of gravity, determine the other image area of the copy In the protruding part. The βχ pixel point is the hand 11. As described in item 10 of the scope of the patent application, the method of identifying and tracking out parts includes the angle between the direction of the line connecting the projection and the center of gravity and the horizontal line based on the second image. The cool angle between the direction of the line connecting the protruding part with the center of gravity and the horizontal line = the threshold value of the protruding angle in the middle; ^: The video sends a trigger message. The distance limit value, 12. As described in item 10 of the scope of patent application, the image is used to identify and track a certain part, wherein the protruding part is the part indicated by * ,,, and ". 1 〇213.879TnV (nl); STLC-01.D-9118; YIANHOU.ptd Page 15