KR101490521B1 - Method for real-time scene-change detection for rate control of video encoder, method for enhancing qulity of video telecommunication using the same, and system for the video telecommunication - Google Patents

Abstract

A real-time scene change detection method according to the present invention is a real-time scene change detection method for controlling a motion-encoded data rate, the method comprising: dividing a current frame into a plurality of regions, and determining a dissimilarity metric (DM) Determining whether the non-similarity measure of each of the regions is out of a preset reference value, confirming the number of regions in the current frame where the non-similarity measure deviates from the preset reference value, It is determined that the scene change has been performed in the current frame when the number of regions out of the reference value is greater than or equal to a predetermined number.

Video, transitions, encoding, real-time, PSNR, prediction

Description

TECHNICAL FIELD [0001] The present invention relates to a real-time scene change detection method, a video call quality improvement method, and a video call system for controlling a video encoded data rate, THE SAME, AND SYSTEM FOR THE VIDEO TELECOMMUNICATION}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving image encoding method and, more particularly, to a real-time scene change detection method for controlling a moving image encoding data rate in moving image encoding.

Various digital moving picture compression techniques have been proposed for obtaining a low data rate or a small storage area while maintaining a high image quality at the time of transmitting or storing moving picture signals. These video compression techniques are international standard standards such as H.261, H.263, H.264, MPEG-2, and MPEG-4. This compression technique achieves a relatively high compression ratio by a discrete cosine transform (DCT) technique or a motion compensation (MC) technique. Such moving picture compression techniques are being applied so that a stream of moving picture data is efficiently transmitted to various digital networks, such as a cellular telephone network, a computer network, a cable network, and a satellite network. And is also efficiently applied to a storage medium such as a hard disk, an optical disk, and a digital moving picture disk (DVD).

For high picture quality, a large amount of data is required in coding a moving picture. However, a communication network carrying moving picture data may limit the data rate that can be applied to coding. For example, a data channel of a satellite broadcasting system or a data channel of a digital cable television network generally transmits data at a constant bit rate (CBR). The storage capacity of storage media such as disks is also limited.

Therefore, the moving image encoding process properly trades off the image quality and the number of bits required for image compression. Also, since moving picture coding requires a relatively complicated process, for example, when the software implements it, the moving picture coding process requires a relatively large number of CPU cycles. Moreover, if it is tried to reproduce it in real time processing, time constraints limit the accuracy in encoding, and thus the achievable image quality is limited.

As described above, the video coding data rate control is an important issue in an actual use environment, and a video coding data rate control scheme for obtaining a high picture quality while reducing the complexity and the transmission data rate of the processing method is proposed.

JVT (Joint Video Team: ITU-T Video Coding Experts Group and ISO / IEC 14496-10 AVC Moving Picture Experts Group, ZG Li, F. Pan, KP Lim, G. Feng, X. Lin, and S. Rahardja, Adaptive basic unit layer rate control for JVT ", JVT-G012-r1, 7th Meeting Pattaya Ⅱ, Thailand, Mar. 2003.) adjusted the quantization parameter (QP) A basic technique for controlling the data rate is disclosed.

On the other hand, when a scene change occurs in an inter frame in a GOP (group of picture) in a video encoding state in a state where a given resource (rate, etc.) is limited, the flow of the control of the encoded data rate is broken. This is because the coding data rate control is made on the assumption that there is similarity with the preceding frame. Real-time scene change detection is required to prevent this situation in advance.

For scene change detection, correlation, statistical sequential analysis, and histogram are used to find similarities between neighboring frames. In the video compressed by H.264 / AVC, an intra-coded macroblock may exist in the inter frame during the Rate Distortion Optimization (RDO) process, and intra-coded intra- If the number of macroblocks exceeds a certain level, the frame can be regarded as a scene change.

However, the method of determining the scene change by the number of intra-coded macroblocks in an interframe in a video compressed with H.264 / AVC is simple, but can not perform real-time processing. That is, the number of intra-coded macroblocks in the interframe can not be known without the quantization parameter (QP) due to the "Chicken & Egg dilemma" occurring in the H.264 / AVC RDO process.

To solve this problem, a method of judging scene change by measuring non-similarity between frames has been studied. The measurement of non-similarity between frames is performed by using a dissimilarity metric (DM) of a compressed image and by using a non-similarity measure (DM) in an uncompressed image.

Since scene change detection is performed for video bit rate control, it must be completed before performing video bit rate control. In addition, the quantization parameter must be calculated before the image compression process through the video bit rate control. As a result, it is not appropriate to calculate the non-similarity measure (DM) in the compressed image since the scene change detection must be performed before performing the image compression.

On the other hand, the non-similarity measure (DM) can be measured using the mean square error (MSE) of the frame in the uncompressed image. Since the computation of the non-similarity measure (DM) using the mean squared error is performed based on the pixels of the frame, there is a disadvantage in that the performance for detecting the scene change is not excellent in a motion image having a large amount of computation . In order to solve this disadvantage, a method of calculating the non-similarity measure (DM) is used in consideration of both the pixels of the frame and the histogram. The four methods of dissimilarity measure (MAFD), MAFD after normalization (HEN), signed difference MAFD (HMA), and HEN (absolute difference frame variance) 4DMs) have been tried. Such a method using 4DMs has excellent detection performance of scene change but a large amount of computation. Therefore, it is not easy to detect the scene change of the frame in real time through the method using 4DMs.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a real-time scene change detection method for controlling a motion-encoded data rate for detecting scene change more efficiently and in real time with less hardware complexity.

It is another object of the present invention to provide a method for detecting an image generated due to an error and improving the quality of the image using the detected image.

According to an aspect of the present invention, there is provided a real-time scene change detection method for real-time scene change detection for controlling a moving picture encoding data rate, the method comprising: dividing a current frame into a plurality of areas, Calculating a dissimilarity metric (DM), determining whether the non-similarity measure of each region is out of a preset reference value, determining whether the non-similarity measure is within a predetermined range, And determining that the scene change has occurred in the current frame when the number of regions out of the reference value is greater than a predetermined number as a result of checking the number of regions that deviate from a preset reference value in the process.

The process of calculating the dissimilarity measure DM of each of the divided regions may be performed by using the error information between samples of the current frame and the reconstructed previous frame (reference frame) to calculate a Peak Signal to Noise Ratio (PSNR) And a process for predicting the probability of occurrence.

The process of calculating the dissimilarity measure DM of each of the divided areas may include calculating a Prediction Peak Signal to Noise Ratio (PSNR) (PPSNR) predicted in the current frame and an average PPSNR To calculate the non-similarity measure of each divided area.

Further, a scene change detection method according to an aspect of the present invention includes the steps of: calculating a differential value of a predicted PSNR of a frame input after a scene change frame; And setting a frame to which a scene change has been completed if the value indicates a negative value.

According to another aspect of the present invention, there is provided a method for video call quality improvement in a wireless terminal, the method comprising: detecting a start frame and an end frame of an abrupt change period of an input video signal of the terminal; The method comprising the steps of: skipping encoding in a transmitter; and copying all frames received in advance in the receiver for the skipped frames and reproducing the skipped frames in place of the skipped frames, And calculates the differential value of the predicted PSNR obtained between the previous images.

A video call system according to another aspect of the present invention is a video call system using a wireless terminal. The video call system includes a detector for detecting a start frame and an end frame of an abrupt change period of an input moving picture signal of the terminal, And a receiver for copying and reproducing the entire frames previously received for the skipped frames, wherein the detector calculates a differential of the predicted PSNR obtained between the input video and the reconstructed previous video Lt; / RTI >

As described above, the real-time scene change detection method for controlling the motion-encoded data rate according to the present invention has less hardware complexity and can detect scene change more efficiently in real time.

In addition, according to the moving picture coding method and system of the present invention, the bit quality of the unfocused image is accumulated and then allocated to the input image, thereby improving the image quality.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. It will be appreciated that those skilled in the art will readily observe that certain changes in form and detail may be made therein without departing from the spirit and scope of the present invention as defined by the appended claims. To those of ordinary skill in the art.

1 is a block diagram of a video encoder apparatus to which a scene change detection method according to an embodiment of the present invention is applied. Referring to FIG. 1, a video encoder apparatus to which a scene change detection method according to an exemplary embodiment of the present invention is applied includes a general H.264 / AVC (Advanced Video Coding ) Encoder 10 as shown in FIG. A frame storage memory 20 for storing frames and an encoder QP controller 30 for performing a quantization parameter (QP) control operation for controlling the data rate of the encoder 10.

The encoder 10 includes a frequency converter 104, a quantizer 106, an entropy coder 108, an encoder buffer 110, an inverse quantizer 116 An inverse frequency transformer 114, a motion estimator / compensator 120, and a filter 112. [

If the current frame is an interframe (e. G., A P frame), the motion estimator / compensator 120 determines whether the current frame is a current frame The motion of the macroblock is estimated and compensated. The frame is processed in units of macroblocks corresponding to, for example, 16x16 pixels of the original image. Each macroblock is coded in an intra or inter mode. When motion estimation is performed, motion information such as a motion vector is output as additional information, and motion compensation is applied to the reconstructed previous frame to generate a motion compensated current frame. The difference between the motion compensated current frame's macroblock (predictive macroblock) and the original current frame's macroblock is provided to the frequency converter 104.

The frequency converter 104 converts the spatial information of the spatial domain into frequency domain (e.g., spectral) data. At this time, the frequency converter 104 typically performs a discrete cosine transform (DCT) to generate a DCT coefficient block on a macroblock-by-macroblock basis.

The quantizer 106 quantizes a block of spectral data coefficients output from the frequency converter 104. At this time, the quantizer 106 applies a constant scalar quantization to the spectral data in a step-size variable on a frame-by-frame basis. The quantizer 106 receives variable quantization parameter (QP) information for each frame from the QP adjuster 34 of the encoder QP controller 30 for data rate control.

The entropy coder 108 compresses the output from the quantizer 106, including specific additional information (e.g., motion information, spatial extrapolation mode, and quantization parameters) of the macroblock. Typical entropy coding techniques include arithmetic coding, Huffman coding, run-length coding, Lempel Ziv coding, and the like. Entropy coder 108 typically applies different coding techniques to different kinds of information.

The moving picture information compressed by the entropy coder 108 is buffered in the encoder buffer 110. The buffer level indicator of the encoder buffer 110 is provided to the encoder QP controller 30 for data rate adjustment. The moving picture information stored in the encoder buffer 110 is output and erased from the encoder buffer 110 at a fixed transmission rate, for example.

On the other hand, when the reconstructed current frame is required for subsequent motion estimation / compensation, the dequantizer 116 performs inverse quantization on the quantized spectral coefficients. The inverse frequency transformer 114 inversely performs the operation of the frequency transformer 104 to generate inverse difference macroblocks from the output of the inverse quantizer 116 through, for example, inverse DCT transform. This is not the same as the original difference macroblock due to the influence of signal loss and the like.

If the current frame is an interframe, the reconstructed inverse difference macroblock is combined with the predictive macroblock of the motion estimator / compensator 120 to generate a reconstructed macroblock. The reconstructed macroblocks are stored in the frame store memory 20 as a reference frame for use in predicting the next frame. In this case, since the reconstructed macroblock is a distorted version of the original macroblock, in some embodiments, the deblocking filter 112 is applied to the reconstructed frame to smooth discontinuity between macroblocks.

The encoder QP controller 30 for controlling the QP of the encoder 10 is provided with a scene change detector 30 for detecting a scene change in real time through a current frame and a reference frame stored in the frame store memory 20 according to a feature of the present invention, (32). When the scene change detection unit 32 detects the scene change, information on the scene change is provided to the QP adjustment unit 34. The QP adjustment unit 34 accordingly adjusts the quantization parameter of the quantizer 106 appropriately So that it can appropriately respond to the scene change of the current frame.

For this, in order to prevent an increase in the calculation load in the process of determining the scene change, the scene change detector 32 determines a value (PPSNR) of the current frame, which is a predicted PSNR (Peak Signal to Noise Ratio). Predicted PSNR). Specifically, the scene change detection unit 32 divides the current frame into a plurality of areas as shown in FIG. 2, and predicts PSNRs of the divided areas. Then, a DM (dissimilarity metric) of each region is calculated, and it is confirmed whether or not the non-affinity measure DM deviates from a predetermined reference value, and whether several regions within the frame deviate from a predetermined reference value Check. If the determined number is greater than or equal to a predetermined value, it is regarded that a scene change has occurred in the current frame.

In the present invention, the dissimilarity measure (DM) of each region is made by checking the ratio of the PSNR of the current frame to the average PPSNR of the previous frames so that local changes in the frame can be confirmed. This can be calculated as shown in Equation (1) below.

In Equation (1), DM ^x _{proposed , where i} denotes an identification number of the divided area, i denotes a frame number of the current frame, S _j denotes a frame number of the corresponding image for the j- . DM ^x _{proposed , i} is the ratio of the PPSNR of the current frame to the average PPSNR of each region from when the frame is changed. Also, PPSNR ^x _{k , k-1} and PPSNR ^x _{i , i-1} can be obtained by the following equations (2) and (3), respectively.

In Equations (2) and (3), n represents the number of bits per sample (i.e., pixel), and generally n can be set to 8. The PMSE is a predicted MSE (Mean Square Error) of the current frame, and can be obtained by the following equations (4) and (5).

In Equations (4) and (5), O ^k _mn and O ⁱ _mn represent original samples of the mth column and the nth row of the kth and ith frames (i.e., the current frame), respectively. One frame consists of M [m] x N [n] pixels. In Equations (4) and (5), O ⁱ _mn denotes an original sample in the mth row and nth row of the i-th frame (i.e., the current frame) and R ^j _mn denotes an m- Represents a reconstructed reference sample. A frame consists of M [m] x N [n] pixels.

In the present invention, whether or not the current frame is switched is determined by checking how many areas having a non-similarity measure (DM ^x _{proposed , i} ) within a predetermined threshold among a plurality of areas constituting a frame.

An area having a non-similarity measure (DM ^x _{proposed , i} ) within a predetermined threshold value is calculated by the following Equation (6), and the scene change can be determined through an operation of Equation (7).

In Equation (6) _{,? Is} a threshold defining the non-similarity measure (DM ^x _{proposed , i} ) of each region. Further, in Equation 7 α is defined by the ratio to determine whether the scene change threshold, the frame, N _f is the number of the divided regions in the frame.

For example, N _f is defined as 12, an alpha value is defined as 0.75, and a beta value is defined as 0.7. At this time, if the number of divided regions having a non-similarity measure (DM ^x _{proposed , i} ) of less than 0.7 is 9 or more, the current frame is judged as a class-switched frame. Further,? And? May be values determined by an experimental value.

3 is a flowchart of a real-time scene detection operation according to an embodiment of the present invention, which is an operation that can be performed in the scene change detection unit 32 as shown in FIG. Referring to FIG. 3, first, an image frame is received through step 302. Next, the input image frame is divided into N _f number of regions through step 304. Then, the non-similarity measure (DM ^x _{proposed , i} ) of each divided region is calculated by calculating Equations (1) to (5) (Step 306). Step 308 compares the non-similarity measure (DM ^x _{proposed , i} ) calculated using Equation (6) with a predetermined threshold value, and calculates C ^x of each region. For example, when the β value is set to 0.7, the calculated dissimilarity measure (DM ^x _{proposed, i)} if relatively small than that of the β value of 0.7 and determining the C ^x 1, the dissimilarity measure (DM ^x _{proposed, i)} β is the value greater than or equal to 0.7 relative to determine the C ^x to zero. And then proceed to step 308, determine the value of x and the value N _f to see the C ^x that all operations of the whole area included in the frame (step 310). For example, assuming that a frame is divided into twelve regions as shown in FIG. 2 (that is, when N _f = 12 is set), among the twelve regions, the region where the non-similarity measure (DM ^x _{proposed , i} ) The x value can be set to 0, and finally the x value of the area where the non-similarity measure (DM ^x _{proposed , i} ) is calculated can be set to 11, N _f -1. Accordingly, step 310 may be to determine whether the x value is equal to the N _f -1 value. As a result of checking in step 310, if the calculation of the C ^x value of the entire area included in the frame is not completed (No in step 310), step 311 is performed to update the x value. Then, 306, 308, 310, and 311 are repeatedly performed until all of the non-similarity measures (DM ^x _{proposed , i} ) and C ^x values of the entire region included in the frame are calculated. On the other hand, if in step 310 a result of the check, the operation of the C ^x value of the whole region included in the frame is complete (310- Yes), it performs step 312. Step 312 sums all of the C ^x value of each region calculated in step 308 using equation (7). Next, in step 314, the summed value is compared with a predetermined threshold, and it is determined whether the current frame is a frame in which the scene change is performed. For example, in the case was 2-304 stage frame divided into 12 areas is the α value is set to 0.75, is considered for each region of the current frame if the result value obtained by summing all of the C ^x value is greater than or equal to level 9 transition is to be made. If it is determined in step 314 that the current frame is level shifted (step 314), a scene change detection signal or the like is generated in step 316, and a non-similarity measure (DM ^x _{proposed , i} ) And updates the S _j value used in the operation. The scene change detection signal generated as described above may be provided to the QP adjusting unit 34. The QP adjusting unit 34 accordingly adjusts the quantization parameter of the quantizer 106 at the time of scene change detection. On the other hand, if it is determined in step 314 that the current frame is not level shifted (No in step 314), step 320 is performed. Step 320 is a step of confirming whether the input frame is the last frame of the image. If it is determined that the inputted frame is the last frame of the image (320-YES), the confirmation of whether or not the frame is changed is terminated. If the input of the frame is continued (320-No) The above steps 302 to 318 are repeatedly performed.

Hereinafter, the effectiveness of the scene change detection method according to the present invention will be examined through experiments. First, we selected two arbitrary test images that were fast for the experiment and made it difficult to detect light-shaded scene transitions. Selected images are shown in Table 1 below. The names of the test sequence images are set to 'Worldcup' and 'FF-X2', respectively. The test sequence images consist of 6,843 frames and 7,138 frames, respectively, and consist of 13 and 159 class transition frames.

Sequence Sequence comment Number of frames Number of grade transitions Worldcup Sports highlight 6,843 13 FF-X2 Animation highlight 7,138 159

The method described in Korean Patent Application No. 10-2006-0075856 (hereinafter referred to as the '856 patent) filed by the applicant of the present invention uses MSE DM, 4DMs according to the prior art using the above two test sequence images, , And the method proposed in the present invention, and the errors generated in the scene change detection process are confirmed and described in Table 2 below. In Table 2, 'number of FALSE' indicates the number detected by scene change even though no actual scene change occurred, and 'number of MISS' indicates the number of times that scene change was not detected even though actual scene change occurred . In addition, DP _FalseMiss (%) was calculated and _{reported for} each method. DP _FalseMiss (%) represents the ratio of FALSE and MISS to the number of scene transitions included in the image, and is a result of the following expression (8).

Sequence ASC detect algorithms Number of FALSE Number of MISS DP _FalseMiss (%) Worldcup MSE DM 2 2 30.8 4DMs 0 One 7.7 The method described in the '856 patent 3 3 46.2 The method described in the present invention 0 One 7.7 FF-X2 MSE DM 97 60 98.7 4DMs 11 50 38.4 The method described in the '856 patent 52 59 69.8 The method described in the present invention 31 48 49.7

Referring to Table 2, when the scene change is detected using the method according to the present invention, the detection performance is relatively higher by about 36.1% than the MSD DM and about 5.7% by about 4% have.

In addition, the computational load required to perform the above-described methods is confirmed through a personal computer, and the results are shown in the graph of FIG.

The personal computer used in the experiment includes Microsoft® Windows® XP as the operating system (OS) and storage media containing Intel VTune ™ Performance Analyzer 8.0 program. The experiment was set to a time-based mode using an operating system timer of the operating system in the personal computer, and the sampling interval was set to 1 ms. In order to increase the reliability of the experiment, three different computers having the above conditions were used and the measured values were averaged using three computers. Figure 4 shows the measured results in terms of the computational load between the algorithms, summing all the timer samples obtained from the three computers. The algorithm proposed by the present invention is improved by 34.8% in comparison with the 'MSE DM' algorithm in terms of the computational load, and is improved by 93.1% over the '4 DMs' algorithm. This is important when considering the computational load of H.264 frame layer rate control as shown in FIG. As a result, the present invention can be applied to an adaptive bit rate control algorithm suitable for class transition in H.264 video coding. In addition, the present invention may be the most suitable algorithm considering the detection performance and computation load as compared with the existing algorithm.

The scene change detection method according to an embodiment of the present invention can be applied to a video call method of a wireless terminal. That is, among the images generated for the video call, the scene change detection method according to an embodiment of the present invention can be appropriately adjusted, encoded, and transmitted with respect to a frame in which scene change has occurred.

On the other hand, along with the physical lens limitation of the mobile device, sudden movement of the mobile device produces an unfocused image. Such an unfocused image consumes a lot of bit resources at the time of coding because there is little spatio temporal correlation. Temporarily consuming a large amount of bit resources can affect the quality of the generated image after a sudden move, which causes the overall image quality to deteriorate. That is, there is a need to improve the quality of images that are not in focus because they allocate more than necessary bits to images that are difficult to view, so that they can be used later.

Accordingly, a method of detecting a scene change according to another embodiment of the present invention proposes a method of detecting frames included in an unfocused image. Specifically, the method for detecting a scene change according to another embodiment of the present invention includes detecting a start frame (hereinafter referred to as a start frame) and a frame ending an unfocused image (Hereinafter referred to as the end frame).

5 is a block diagram of a moving picture encoder apparatus to which a scene change detection method according to another embodiment of the present invention is applied. Referring to FIG. 5, the moving picture encoder apparatus to which the scene change detection method according to another embodiment of the present invention is applied is the same as the moving picture encoder apparatus to which the scene change detection method according to an embodiment of the present invention is applied. However, the detailed configuration of the scene change detection unit 32 is different. Therefore, in the moving picture encoder apparatus of another embodiment of the present invention, the same reference numerals are given to the same constituent elements as those of the moving picture encoder apparatus of one embodiment. A detailed description of the same components as those of the moving picture encoder apparatus of one embodiment will be omitted because it is disclosed in detail in one embodiment of the present invention.

In the moving picture encoder apparatus according to another embodiment of the present invention, the encoder controller 40 for controlling the encoder 10 is provided with a current frame and a reference frame stored in the frame storage memory 20 according to a feature of the present invention And a scene change detector 42 for detecting a very fast scene in real time. The scene change detection unit 42 includes a start frame detection unit 422 for detecting a start frame of a similar scene, an end frame detection unit 424 for detecting an end frame of a similar scene, a frame skip determination unit 426 for determining a frame to be skipped, Respectively.

When the scene change detection unit 42 detects the scene change, information on the scene change is provided to the QP adjustment unit 44. The QP adjustment unit 44 accordingly adjusts the quantization parameters of the quantizer 106 appropriately So that it can appropriately respond to the scene change of the current frame.

The start frame detector 422 predicts a Peak Signal to Noise Ratio (PSNR) of a reference frame and an input current frame to determine a scene change. That is, when the predicted PSNR deviates from a preset reference value, it is considered that the scene change has occurred in the current frame.

Furthermore, the start frame detecting unit 422 may detect the start frame through the same operation as the scene change detecting unit 32 in the embodiment of the present invention.

The end frame detector 322 detects the end frame using a differential value of a parameter that predicts a reference frame and a predicted PSNR (Peak Signal to Noise Ratio) of the current frame. For example, the end frame detector 322 calculates the following equation (9) to detect the end frame.

That is, the negative Diff _{AvgPartialPPSNR} from the viewpoint of the motion of the image means that the motion of the current frame is _smaller than the motion of the previous frame. On the basis of this, the end frame detector 424 detects a very fast image (i.e., an out-of-focus image) in the start frame detector 422, and then ends the frame in which the Diff _{AvgPartialPPSNR} value becomes negative for the first time Frame.

In Equation (9), PPSNR is a parameter that predicts a stored reference frame and a PSNR of an input current frame, and can be calculated as Equation (3) in an embodiment of the present invention. N _f indicates the number of blocks into which the frame is divided.

On the other hand, the skip determiner 426 determines a frame to be skipped from the information of the start frame detector 422 and the end frame detector 424. If the frame to be omitted is determined, the compressed data of the frame is not transmitted, and only the information indicating that the frame is omitted is transmitted to the entropy coder 108.

The QP adjuster 34 receives information on the end frame of the very fast video (i.e., out of focus) from the end frame detector 424. The QP adjusting unit 34 performs a data rate control for applying a quantization parameter according to the complexity of an image to a frame input after a very fast image (that is, an out of focus image) ends .

Furthermore, the scene change detection method according to another embodiment of the present invention can be applied to a video call method of a wireless terminal. That is, it is possible to detect a very fast image (for example, an out-of-focus image) using the scene change detection method according to another embodiment of the present invention, skip the corresponding frame and transmit it. In addition, the receiving end can restore the previous frame instead of the skipped image.

Specifically, a video call method of a wireless terminal can be performed by a wireless terminal having a video transmission apparatus and a video receiving apparatus.

An image transmission apparatus provided in a wireless terminal encodes an image photographed by a camera provided for a video call during a video call and transmits the encoded image to a video receiving apparatus. At this time, the image transmission apparatus detects a frame in which a very fast image (for example, an out of focus image) starts, and detects a frame in which the very fast image ends, out of images photographed by the camera. In particular, the image transmission device can detect a frame in which a very fast image (e.g., out of focus) starts, in a manner disclosed in a scene change detection method according to an embodiment of the present invention, Can be detected by differential calculation of PPSNR. Further, the image transmission apparatus performs coding by inserting a signal indicating skipping of a frame, which exists between a frame in which a very fast image (for example, a non-focused image) starts and a frame in which it ends, Lt; / RTI >

On the other hand, the video receiving apparatus recovers video data and reproduces the restored video through a display device. The video receiving apparatus can confirm a signal indicating an inserted frame skip in the encoding process, thereby copying a frame that exists immediately before the skipped frame and a skipped frame, replacing the skipped frame, .

The method according to the present invention can be implemented as computer readable code on a computer readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical disk and the like, and also a carrier wave (for example, transmission via the Internet) do. The computer-readable recording medium may also be distributed over a networked computer system and stored and executed in computer readable code in a distributed manner.

As described above, the real-time scene change detection operation for controlling the coding rate of the moving image according to the embodiment of the present invention can be performed. While the present invention has been described with respect to the specific embodiments, Can be carried out without departing from the scope.

FIG. 1 is a block diagram of a moving picture encoder apparatus to which a scene change detection method according to an embodiment of the present invention is applied.

Figure 2 is an example of a frame divided into a plurality of regions according to an embodiment of the present invention,

3 is a flowchart of a real-time scene detection operation according to an embodiment of the present invention,

4 is a graph showing test results of a real-time scene detection operation according to an embodiment of the present invention,

5 is a block diagram of a moving picture encoder apparatus to which a scene change detection method according to another embodiment of the present invention is applied.

Claims (10)

A real-time scene change detection method for controlling a coded video data rate, Dividing a current frame into a plurality of regions and computing a dissimilarity metric (DM) of each of the divided regions; Determining whether the non-similarity measure of each of the regions is out of a preset reference value; Confirming the number of regions within the current frame where the non-similarity measure deviates from the preset reference value; And determining that a scene change has occurred in the current frame when the number of regions out of the reference value is greater than a predetermined number as a result of checking the number of regions that deviate from a preset reference value in the process, Way. 2. The method of claim 1, wherein calculating the dissimilarity measure (DM) of each of the divided regions comprises: Estimating a PSNR (Peak Signal to Noise Ratio) of a current frame before encoding using error information between samples of a current frame and a reconstructed previous frame (reference frame) . 3. The method according to claim 2, wherein the step of calculating the dissimilarity measure (DM) Similarity measure of each divided region is calculated by using the predicted Peak Signal to Noise Ratio (PSNR) (hereinafter referred to as PPSNR) predicted in the current frame and the average PPSNR of the frames after the scene change has occurred Time scene transition detection method. 3. The real-time scene change detection method according to claim 2, wherein the non-similarity measure of each divided region is calculated according to Equation (1) below. &Quot; (1) "

x denotes the identification number of the divided area, i denotes the frame number of the current frame, and S _j denotes the frame number of the corresponding image for the j-th occurrence of class transition. 5. The method of claim 4, wherein the PPSNR is calculated according to Equation (2) and Equation (3). &Quot; (2) "

&Quot; (3) "

In Equations (2) and (3), PMSE is a predicted MSE (Mean Square Error) of the current frame, and n is the number of bits per sample. The PMSE _{i, i-1} and PMSE _{k, k-1} are calculated according to the following equations (4) and (5), respectively. &Quot; (4) "

Equation (5)

In Equations (4) and (5), O ⁱ _mn denotes an original sample in the mth column and nth row of the ith frame, and R ^j _mn denotes a reconstructed sample in the mth row and nth row in the jth frame (One frame consists of M [m] x N [n] pixels). 2. The method of claim 1, wherein the determination of whether or not the number of regions out of the reference value is equal to or greater than a predetermined number, Is determined by an operation according to Equation (6) below. &Quot; (6) "

In Equation (6),? Is a threshold value defining a ratio for determining whether or not the scene is switched, and N _f is the number of regions divided in the frame. In Equation (6), C ^x is determined by an operation according to Equation (7) below. &Quot; (7) "

In Equation (7),? Is a threshold value defining the non-similarity measure of each region. 7. The method according to any one of claims 1 to 6, Calculating a differential value of a predictive PSNR of a frame input after the scene change frame; Further comprising the step of checking the calculated result value and setting the resultant value as a frame in which the scene change has been completed if the result value indicates a negative value. 8. The method of claim 7, wherein the differential value of the prediction PSNR is calculated using Equation (8). &Quot; (8) "

In the equation (8) Prediction PSNR is a parameter which predicts the PSNR of a current frame being input and stored in the reference frame (reference frame), hereinafter N _f denotes the frame number of the divided blocks. A video call method of a wireless terminal, Detecting a start frame and an end frame of an abrupt change period of an input moving picture signal of the terminal, Wherein the transmitting unit skips the encoding for the detected images, And a step of copying the previous frame received in advance in the receiver for the skipped frames and playing back the skipped frame in place of the skipped frame, Wherein the start frame is verified using a non-similarity measure between respective regions of a plurality of divided frames, and the end frame is confirmed through a differential of a predictive PSNR obtained between an input image and a reconstructed previous image How to improve call quality. A video call system using a wireless terminal, A detector for detecting a start frame and an end frame of an abrupt change period of an input moving picture signal of the terminal, A transmission unit for skipping frame encoding for all detected images, And a reception unit for copying and playing back the previous frames previously received for the skipped frames, The detector identifies the start frame using a non-similarity measure between regions of a plurality of divided frames, and identifies the end frame through a differential of a predictive PSNR obtained between an input image and a reconstructed previous image Video call system.

Images