KR20090037288A - 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 for controlling a moving picture encoding data rate, a method for improving video call quality through the same and a video call system are provided to assign bit resources of an image which is not focused to a next input image, thereby improving image quality. A frequency converter(104) performs discrete cosine transform. A quantizer(106) quantizes a block of a spectrum data coefficient outputted in the frequency converter. The quantizer applies predetermined scalar quantization to spectrum data by a step size. The quantizer receives variable information of quantization parameter per a frame from a QP adjusting unit(34). An entropy coder(108) compresses specific additional information of a corresponding macro block and output from of the quantizer. Moving picture information compressed in an entropy coder is buffered in an encoder buffer(110). A buffer level pointer of the encoder buffer is provided to an encoder QP controller(30).

Description

METHODE 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}

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to video encoding, and relates to a method for detecting real-time scene changeover that is preceded for video coded data rate control during video encoding.

Various digital video compression techniques have been proposed to obtain a low data rate or a small storage area while maintaining high image quality in transmission or storage of a video signal. These video compression technologies are international standards such as H.261, H.263, H.264, MPEG-2, MPEG-4, and the like. Such a compression technique achieves a relatively high compression ratio by a Discrete Cosine Transform (DCT) technique or a Motion Compensation (MC) technique. Such video compression technology is applied to efficiently stream streams of video data to various digital networks, for example, cellular networks, computer networks, cable networks, satellite networks, and the like. It is also applied to storage media such as hard disks, optical disks, and digital video disks (DVD).

For high quality, a large amount of data is required for video encoding. However, a communication network that delivers video data may limit the data rate applicable to encoding. For example, data channels of satellite broadcasting systems or data channels of digital cable television networks generally transmit data at a constant bit rate (CBR). In addition, the storage capacity of a storage medium such as a disk is also limited.

Therefore, the video encoding process trades off the image quality and the number of bits necessary for image compression. Also, since video encoding requires a relatively complicated process, for example, to implement it in software, the video encoding process requires a relatively large number of CPU cycles. Moreover, when attempting to reproduce this by real time processing, time constraints limit the precision in performing encoding, thereby limiting the image quality that can be achieved.

As described above, video coded data rate control is an important issue in an actual use environment, and a video coded data rate control method has been proposed to obtain high quality while reducing the complexity of the processing method and the transmission data rate.

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.) adjusts the quantization parameter (QP) in video frame coding according to MPEG video compression algorithm. Basic techniques for controlling the data rate have been disclosed.

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

For scene change detection, a technique for finding similarity between neighboring frames uses a method such as correlation, statistical sequential analysis, and histogram. In addition, in an H.264 / AVC compressed video, intra-coded macroblocks may exist in an inter frame during a rate distortion optimization (RDO) process, and intra-coded in an inter frame. If the number of macroblocks is above a certain level, the frame can be considered to be a transition.

However, in the H.264 / AVC-compressed video, the method of determining a scene change by the number of intra-coded macroblocks in an inter frame is simple but cannot perform real-time processing. That is, the number of macroblocks intra-coded in an inter frame without a quantization parameter (QP) may not be known by “Chicken & Egg dilemma” generated in the H.264 / AVC RDO process.

In order to solve this problem, a method of determining a scene transition by measuring dissimilarities between frames has been studied. The measurement of dissimilarity between frames includes a method of using a dissimilarity metric (DM) of a compressed image and a method of using a dissimilarity measure (DM) of an uncompressed image.

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

Meanwhile, in the uncompressed image, the dissimilarity measure (DM) may be measured using a mean square error (MSE) of the frame. Computing the dissimilarity measure (DM) using the mean square error is not very expensive because the calculation is performed based on the pixels of the frame, but there is a disadvantage in that the performance of detecting a scene change in a moving image is not excellent. . To alleviate this drawback, a method of calculating dissimilarity measures (DM) using both the pixel and histogram of the frame is used. As described above, four non-similarity measures (mean absolute frame difference (MAFD), MAFD after histogram equalization with normalization (HEN), signed difference MAFD (SDMAFD) after HEN, absolute difference frame variance (ADFV) after HEN) 4DMs) have been tried. The method using 4DMs has a good detection performance of scene switching but a large amount of calculation. Therefore, it is not easy to detect a scene change of a frame in real time through the method using 4DMs.

SUMMARY OF THE INVENTION The present invention has been made in view of the foregoing, and an object thereof is to provide a real-time scene change detection method for controlling video coded data rate for detecting scene change in real time with less hardware complexity.

Another object of the present invention is to provide a method of detecting an image generated due to an error and improving the quality of the image by using the same.

In order to achieve the above object, the real-time scene change detection method according to the present invention is a real-time scene change detection method for controlling video encoding data rate, wherein the current frame is divided into a plurality of areas, and the similarity of each divided area is not similar. Calculating a dissimilarity metric (DM), determining whether a non-similarity measure of each region deviates from a preset reference value, and a number of regions in which a non-similarity measure deviates from the preset reference value within a current frame And checking the number of regions that deviate from the preset reference value in the process, and deciding that a scene change is made in the current frame when the number of regions deviating from the reference value is greater than or equal to the preset number.

The process of calculating the dissimilarity measure (DM) of each divided region may include a peak signal to noise ratio (PSNR) of a current frame by using error information between samples between a current frame and a reconstructed previous frame (reference frame). It can include the process of predicting.

The process of calculating the dissimilarity measure (DM) of each divided region includes a predicted peak signal to noise ratio (PSNR) predicted in the current frame and an average PPSNR of frames after the scene change occurs. Using the ratio of and to calculate the similarity measure of each divided region.

Furthermore, the scene change detection method according to an aspect of the present invention, the process of calculating the derivative value of the predicted PSNR of the frame input after the frame, the scene change is made, and confirming the calculated result value, When indicating a negative value, the method may further include setting the frame to which the scene change is completed.

According to another aspect of the present invention, there is provided a method for improving a video call quality in a video call method of a wireless terminal, the method comprising: detecting a start frame and an end frame of a sudden change section of an input video signal of a terminal; The transmitting unit skips encoding, and the receiving unit copies the entire frame received in advance and reproduces the skipped frame instead of the skipped frames. The end frame detection is performed by reconstructing the input image. It calculates through the derivative of the predictive PSNR obtained between the previous images.

According to another aspect of the present invention, a video call system includes a detector for detecting a start frame and an end frame of a sudden change section of an input video signal of a terminal in a video call system using a wireless terminal. And a transmitter for skipping frame encoding for a receiver, and a receiver for copying and reproducing a previous frame previously received for the skipped frames, wherein the detector is a derivative of a predicted PSNR obtained between an input image and a reconstructed previous image. Calculate through

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

In addition, according to the video coding method and system according to the present invention, it is possible to improve the image quality by accumulating the bit resources of the out of focus image and assigning them to the input image.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, specific details such as specific components are shown, which are provided to help a more general understanding of the present invention, and it is understood that these specific details may be changed or changed within the scope of the present invention. It is self-evident to those of ordinary knowledge in Esau.

1 is a block diagram of a video encoder device 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 receives a video frame sequence and outputs compressed video data as an output, which is general H.264 / AVC (Advanced Video Coding). The encoder 10 may be provided. Also, a frame storage memory 20 for storing frames and an encoder QP controller 30 for performing a quantization parameter (QP) control operation for data rate control of the encoder 10 are provided.

First, the configuration and operation of the encoder 10 will be described in more detail. The encoder 10 includes a frequency converter 104, a quantizer 106, an entropy coder 108, an encoder buffer 110, and an inverse quantizer 116. ), An inverse frequency converter 114, a motion estimation / compensator 120, and a filter 112.

If the current frame is an inter frame (e.g., a P frame), the motion estimation / compensator 120 may determine the current frame within the current frame, relative to the reference frame, which is a reconstructed frame of the previous frame buffered in the frame storage memory 20. I) Estimate and compensate for the motion of the macroblock. The frame is processed in units of macroblocks corresponding to, for example, 16x16 pixels of the original image. Each macroblock is encoded in an intra or inter mode. In motion estimation, motion information such as a motion vector is output as additional information, and in motion compensation, motion information is applied to a reconstructed previous frame to generate a motion compensated current frame. The difference between the macroblock (prediction macroblock) of the motion compensated current frame and the macroblock of the original current frame is provided to the frequency converter 104 in this manner.

The frequency converter 104 converts video information of the spatial domain into frequency domain (eg, spectrum) data. In this case, the frequency converter 104 typically performs Discrete Cosine Transform (DCT) to generate DCT coefficient blocks in macroblock units.

Quantizer 106 quantizes a block of spectral data coefficients output from frequency converter 104. In this case, the quantizer 106 applies a constant scalar quantization to spectral data with a step-size that is usually changed on a frame-by-frame basis. The quantizer 106 receives variable information of the quantization parameter QP for each frame from the QP adjuster 34 of the encoder QP controller 30 for data rate control.

Entropy coder 108 compresses the output from quantizer 106, including certain side information (eg, motion information, spatial interpolation modes, quantization parameters) of the macroblock. Commonly applied entropy coding techniques include arithmetic coding, Huffman coding, run-length coding, Lempel Ziv (LZ) coding, and the like. Entropy coder 108 typically applies different coding techniques to different kinds of information.

The video 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. Video information stored in the encoder buffer 110 is output and deleted from the encoder buffer 110 at a fixed transmission rate, for example.

On the other hand, if 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 converter 114 reversely operates the frequency converter 104 such that an inverse difference macroblock is generated from the output of the inverse quantizer 116, for example, via an inverse DCT transform. This is not the same as the original difference macroblock due to signal loss and the like.

When the current frame is an inter frame, the reconstructed inverse difference macroblock is combined with the prediction macroblock of the motion estimation / compensator 120 to generate a reconstructed macroblock. The reconstructed macroblocks are stored in the frame storage memory 20 as reference frames 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, thereby facilitating discontinuities between macroblocks.

Meanwhile, the encoder QP controller 30 that controls the QP of the encoder 10 includes a scene change detection unit that detects a scene change in real time through a current frame and a reference frame stored in the frame storage memory 20 according to a feature of the present invention. (32) is provided. When the scene change detection unit 32 detects a scene change, information about the scene change is provided to the QP adjusting unit 34, and the QP adjusting unit 34 accordingly adjusts the quantization parameter of the quantizer 106 properly when detecting the scene change. Adjust to adapt to the transition of the current frame.

To this end, in the present invention, the scene change detection unit 32 predicts a peak signal to noise ratio (PSNR) of a current frame (PPSNR) in order to prevent an operation load from increasing during the scene change determination. Use only Predicted PSNR). Specifically, the scene change detector 32 divides the current frame into a plurality of regions as shown in FIG. 2 and predicts the PSNR of each divided region. The dissimilarity metric (DM) of each region is calculated to determine whether the dissimilarity metric deviates from a predetermined reference value, respectively, and how many regions within a frame deviate from a predetermined reference value. Check it. If the checked number is more than a predetermined value, it is considered that a scene change has occurred in the current frame.

In the present invention, the dissimilarity measure (DM) of each region consists of checking the ratio of the PSNR of the current frame and the average PPSNR of previous frames so as to identify local changes in the frame. This may be calculated as in Equation 1 below.

In Equation 1, in DM ^x _{proposed , i} denotes an identification number of a divided region, i denotes a frame number of the current frame, and S _j denotes a frame number of the corresponding image for the j-th occurrence of the sudden switching. Means. DM ^x _{proposed , i} is the ratio between the PPSNR of each region of the current frame and the average PPSNR of each region from the time when the scene change of the frame is made. In addition, PPSNR ^x _{k , k-1} and PPSNR ^x _{i , i-1} may be obtained by Equations 2 and 3, respectively.

In Equations 2 and 3, n denotes the number of bits per sample (ie, a pixel), and in general, n may be set to 8. The PMSE is a predicted Mean Square Error (MSE) of the current frame, and may be obtained by the calculation of Equations 4 and 5 below.

In Equations 4 and 5, O ^k _mn and O ⁱ _mn represent the m-th row n-th original samples of the k-th and i-th frames (that is, the current frame), respectively. One frame consists of M [m] x N [n] pixels. In Equations 4 and 5, O ⁱ _mn represents an m-th row n original sample of the i-th frame (that is, the current frame), and R ^j _mn represents the m-th n-th row of the j-th frame (ie, the previous frame). Reconstructed reference samples are shown. One frame consists of M [m] x N [n] pixels.

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

An area having a dissimilarity measure (DM ^x _{proposed , i} ) within a predetermined threshold may be calculated by Equation 6 below, and whether or not to change scenes may be determined by using Equation 7 below.

Β in Equation 6 is a threshold that defines the non-similarity measure (DM ^x _{proposed , i} ) of each region. In Equation 7, α is a threshold defining a ratio for determining whether a frame is changed or not, and N _f represents the number of regions divided in the frame.

For example, N _f is defined as 12, α value is defined as 0.75, and β value is defined as 0.7. At this time, if the number of partitions having a similarity measure (DM ^x _{proposed , i} ) less than 0.7 is 9 or more, it is determined that the current frame is a suddenly switched frame. Furthermore, α and β may be values determined by experimental values.

3 is a flowchart of a real-time scene detection operation according to an embodiment of the present invention, which may be performed by the scene change detection unit 32 as shown in FIG. 1. Referring to FIG. 3, first, an image frame is received in step 302. Next, the input image frame is divided into a plurality of N _f regions in step 304. In operation 306 _, the non-similarity measure DM ^x _{proposed and i} of each divided region may be calculated by calculating Equations 1 to 5. Step 308 compares the non-similarity measure (DM ^x _{proposed , i} ) calculated using Equation 6 with β value, which is a predetermined threshold, and calculates C ^x of each region. For example, when the β value is set to 0.7, when the calculated similarity measure DM ^x _{proposed , i} is relatively smaller than the β value 0.7, the C ^x is determined as 1, and the similarity measure DM ^x _{proposed , If i} ) is relatively greater than or equal to the β value of 0.7, the C ^x is determined to be zero. After proceeding to step 308, by checking the x value and the N _f value it is confirmed whether all the C ^x of the entire region included in the frame has been calculated (step 310). For example, assuming that a frame is divided into 12 regions as shown in FIG. 2 (that is, when N _f = 12 is set), among the 12 regions, the first non-similarity measure DM ^x _{proposed , i} , etc. is calculated. The x value may be set to 0, and finally, the x value of the region in which the dissimilarity measure DM ^x _{proposed , i} , etc. is calculated may be set to 11, which is 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, when the calculation of the C ^x value of the entire area included in the frame is not completed (310-no), step 311 is performed to update the x value. Then, the perform 306, 308, 310, and step 311 until the dissimilarity measure (DM _{proposed ^x, i)} and C ^x the value of the entire region including all of the frame to be computed repeatedly. On the other hand, if it is determined in step 310 that the calculation of the C ^x value of the entire area included in the frame is completed (310-Yes), step 312 is performed. In operation 312, all of the C ^x values of each region calculated in operation 308 are added up using Equation 7. Next, step 314 compares the summed value with a predetermined threshold to determine whether the current frame is a frame that has undergone a sudden scene change. For example, when the frame is divided into 12 regions in step 304 and the α value is set to 0.75, if the resultant value of the sum of all the C ^x values of each region is 9 or more, the current frame is regarded as a sudden scene change. If it is determined in step 314 that the current frame has undergone a rapid scene change (Yes-314), a scene change detection signal is generated in step 316, and in step 318, the non-similarity measure (DM ^x _{proposed , i} ) Update the S _j value used in the operation. The scene change detection signal generated as described above may be provided to the QP adjuster 34, and the QP adjuster 34 accordingly adjusts the quantization parameter of the quantizer 106 according to the scene change detection. On the other hand, if it is determined in step 314 that the current frame does not have a sudden scene change (314-No), step 320 is performed. Step 320 is to determine whether the input frame is the last frame of the image. If the inputted frame is the last frame of the image (320-Yes) as a result of the check in step 320, the checking of whether the frame is transitioned is terminated, and if the frame is continuously inputted (320-No), the input of the frame is terminated. The above steps are repeated until step 302 to step 318 are performed.

Hereinafter, the effectiveness of the scene change detection method according to the present invention is examined through experiments. First, two random test images were selected for the experiment, which make the movement fast and make it difficult to detect the diversion of the illumination light. The selected image is shown in Table 1 below. The name of the test sequence image is set to 'Worldcup' and 'FF-X2', respectively. The test sequence images consisted of 6,843 and 7,138 frames, respectively, and consisted of 13 and 159 sharp scene switching frames.

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

Method described in Korean Patent Application No. 10-2006-0075856, previously filed by the applicant of the present invention, MSE DM, 4DMs according to the prior art using the two test sequence images described above (hereinafter, the method described in the '856 patent) According to the method proposed in the present invention, and to detect the scene change, respectively, and to check the error generated during the scene change detection process is described in Table 2 below. In Table 2, 'the number of FALSE' indicates the number of scenes detected by the transition even though no actual transition occurred, and the 'number of MISS' indicates the number of scenes not detected even though the actual transition occurred. Indicates. In addition, DP _FalseMiss (%) was calculated and described for each method. DP _FalseMiss (%) represents the ratio of the sum of FALSE and MISS to the number of scene transitions included in the image, which is the result of calculating Equation 8 below.

Sequence ASC detect algoriths Number of FALSE Number of MISS DP _FalseMiss (%) Worldcup MSE DM 2 2 30.8 4DMs 0 One 7.7 Method described in the '856 patent 3 3 46.2 Method described in the present invention 0 One 7.7 FF-X2 MSE DM 97 60 98.7 4DMs 11 50 38.4 Method described in the '856 patent 52 59 69.8 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 superior to about 36.1% relative to MSD DM, and about 5.7% lower than 4DMs. have.

In addition, the computational load required to perform the above-described methods was confirmed through a personal computer, and the result is illustrated through the graph of FIG. 4.

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

Such a scene change detection method according to an embodiment of the present invention may be applied to a video call method of a wireless terminal. That is, among frames of scenes generated for a video call, a scene change occurs, a scene change detection method according to an embodiment of the present invention may be appropriately adjusted, encoded, and transmitted.

On the other hand, in addition to the physical lens limitation of the mobile device, sudden movement of the mobile device generates an out of focus image. Since the defocused video has little space-time correlation, a lot of bit resources are consumed during encoding. Temporary heavy consumption of bit resources affects well-formed images after sudden movements, which in turn degrades the overall image quality. In other words, by assigning more bits than necessary to an image that is difficult to see, it is necessary to improve the image since it affects a proper image later.

Therefore, the image scene change detection method according to another embodiment of the present invention proposes a method that can detect the frames included in the out of focus image. Specifically, the image scene change detection method according to another embodiment of the present invention is a frame in which a scene change starts (hereinafter referred to as a start frame) due to an image that is not in focus and a frame in which an image that is not in focus ends ( Hereinafter referred to as end frame.).

5 is a block diagram of a video encoder device to which a scene change detection method according to another embodiment of the present invention is applied. Referring to FIG. 5, a video encoder device to which a scene change detection method according to another embodiment of the present invention is applied is the same as a video encoder device to which a 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 video encoder device of another embodiment of the present invention, the same components as those of the video encoder device of one embodiment are assigned the same reference numerals. In addition, detailed descriptions of the same components as those of the video encoder apparatus of one embodiment are omitted since they are disclosed in detail in one embodiment of the present invention.

Meanwhile, in the video encoder device to which another embodiment of the present invention is applied, 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 the feature of the present invention. A scene change detector 42 for detecting a very fast scene in real time is provided. The scene change detection unit 42 includes a start frame detector 422 for detecting a start frame of a similar scene, an end frame detector 424 for detecting an end frame of a similar scene, and a frame skip determination unit 426 for determining a frame to be skipped. It is provided.

When the scene change detection unit 42 detects a scene change, information about the scene change is provided to the QP adjusting unit 44, and the QP adjusting unit 44 accordingly adjusts the quantization parameter of the quantizer 106 appropriately when detecting the scene change. Adjust to adapt to the transition of the current frame.

The start frame detector 422 determines a scene change by predicting a peak signal to noise ratio (PSNR) between a previously stored reference frame and an input current frame. That is, when the predicted PSNR deviates from a preset reference value, it is assumed that a scene change occurs in the current frame.

In addition, the start frame detector 422 may detect the start frame through the same operation as the scene change detector 32 according to the exemplary embodiment of the present invention.

The end frame detector 322 detects the end frame by using a derivative value of a parameter predicted a PSNR (Peak Signal to Noise Ratio) of a previously stored reference frame and an input current frame. For example, the end frame detector 322 detects the end frame by calculating Equation 9 below.

That is, from the viewpoint of the motion of the image, the negative Diff _{AvgPartialPPSNR} means that the motion of the current frame is reduced compared to the motion of the previous frame. Based on this, the end frame detector 424 ends the first frame having a negative Diff _{AvgPartialPPSNR} value after a very fast image (that is, an out of focus image) is detected by the start frame detector 422. Decide on a frame.

In Equation 9, PPSNR is a parameter for predicting a stored reference frame and a PSNR of an input current frame, and may be calculated as in Equation 3 of an embodiment of the present invention. N _f indicates the number of blocks in which the frame is divided.

Meanwhile, the skip determination unit 426 determines a frame to be omitted from the information of the start frame detector 422 and the end frame detector 424. If it is determined that the frame is to be omitted, the compressed data of the frame is not transmitted, and only information indicating that the frame is omitted is transmitted to the entropy coder 108.

The QP adjuster 34 receives information about an end frame of a very fast image (that is, an image that is not in focus) from the end frame detector 424. The QP adjuster 34 performs data rate control for applying a quantization parameter according to the complexity of the image to a frame input after a very fast image (that is, an image that is not in focus) is finished. To perform.

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, by using the scene change detection method according to another embodiment of the present invention, it is possible to detect a very fast image (for example, an image that is out of focus) and skip and transmit a frame corresponding thereto. In addition, the receiver may restore the previous frame instead of the skipped image.

In detail, the video call method of the wireless terminal may be performed by a wireless terminal having a video transmitting apparatus and a video receiving apparatus.

The video transmitting apparatus provided in the wireless terminal encodes an image photographed from a camera provided for video calling and transmits the video to the video receiving apparatus during the video call. In this case, the image transmission apparatus detects a frame at which a very fast image (for example, an out of focus image) starts from among images captured by the camera, and detects a frame at which the very fast image ends. In particular, the image transmission apparatus may detect a frame at which a very fast image (eg, an image that is out of focus) starts by the method disclosed in the scene change detection method according to an embodiment of the present invention, and the very fast image is terminated. The frame to be detected can be detected through a derivative operation of the PPSNR. Furthermore, the image transmission apparatus performs encoding by inserting a signal indicating skip of a frame to a frame existing between a frame at which a very fast image (for example, an image that is not in focus) starts and ends the frame. To send.

On the other hand, the image receiving apparatus is encoded to restore the image data, and reproduces the restored image through the display device. The image receiving apparatus may identify a signal indicating skipping of the inserted frame during the encoding process, thereby copying the skipped frame and the frame immediately present in time, and replacing the skipped frame to replace the skipped frame. Will be restored.

The method according to the invention can be embodied as computer readable code on a computer readable recording medium. Computer-readable recording media include all kinds of recording devices that store data that can be read by a computer system. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, and the like, and may also include those implemented in the form of carrier waves (eg, transmission over the Internet). do. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

As described above, a real-time scene change detection operation for controlling video encoding data rate according to an embodiment of the present invention can be performed. Meanwhile, in the above description of the present invention, a specific embodiment has been described. It can be carried out without departing from the scope.

1 is a block diagram of a video encoder device to which a scene change detection method is applied according to an embodiment of the present invention;

2 is an exemplary view of a frame divided into a plurality of areas 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 a test result of a real-time scene detection operation according to an embodiment of the present invention;

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

Claims (10)

In the real-time scene change detection method for controlling the video coded data rate, Dividing a current frame into a plurality of regions and calculating a dissimilarity metric (DM) of each divided region; Determining whether the similarity measure of each region deviates from a preset reference value; Confirming the number of regions within the current frame where the similarity measure deviates from the preset reference value, And checking the number of regions outside the preset reference value in the above process, when the number of regions outside the reference value is greater than or equal to the preset number, determining that the scene has been changed in the current frame. Way. The method of claim 1, wherein the calculating of the similarity measure (DM) of each divided region comprises: A method of predicting a peak signal to noise ratio (PSNR) of a current frame before encoding by using error information between samples between a current frame and a reconstructed previous frame (reference frame) . The method of claim 2, wherein the non-similarity measure (DM) of each divided region is calculated. After the PSNR predicted in the current frame (hereinafter referred to as PPSNR) and the scene transition occurs, the non-similarity measure of each divided region is calculated using the average PPSNR of the frames. Real time scene change detection method. The method according to claim 2, wherein the similarity measure of each divided region is calculated according to Equation 1 below. <Equation 1>

x denotes an identification number of the divided region, i denotes a frame number of the current frame, and S _j denotes a frame number of the corresponding image for the jth sudden scene changeover. The method according to claim 4, wherein the PPSNR is calculated according to Equation 2 and Equation 3 below. <Equation 2>

<Equation 3>

In Equation 2 and Equation 3, PMSE is the predicted Mean Square Error (MSE) of the current frame, and in Equation 10, n is the number of bits per sample. The PMSE _{i , i- 1} and PMSE _{k , k-1} are calculated according to Equations 4 and 5, respectively. <Equation 4>

<Equation 5>

In equations (4) and (5), O ⁱ _mn represents an m-th row n-row original sample of the i-th frame, and R ^j _mn represents the m-th row n-th reconstructed sample of the j-th frame. (One frame consists of M [m] x N [n] pixels). The method of claim 1, wherein the determination of whether the number of regions deviating from the reference value is greater than or equal to a preset number is performed. Real-time scene change detection method characterized in that determined by the operation according to the equation (6). <Equation 6>

In Equation 6, α is a threshold defining a ratio for determining whether a frame is transitioned, and N _f is the number of regions divided in the frame. In Equation 14, C ^x is determined by an operation according to Equation 15 below. <Equation 7>

In Equation 7, β is a threshold defining a non-similarity measure of each region. The method according to any one of claims 1 to 6, Calculating a derivative value of a predictive PSNR of a frame input after the frame on which the scene change is performed; And checking the calculated result value and setting the frame to a frame at which the scene change is completed when the result value indicates a negative value. 8. The method according to claim 7, wherein the derivative value of the predicted PSNR is calculated using Equation 8 below. <Equation 8>

In Equation 8, the prediction PSNR is a parameter for predicting the stored reference frame and the PSNR of the input current frame, and N _f indicates the number of blocks in which the frame is divided. In the video call method of the wireless terminal, Detecting a start frame and an end frame of a sudden change section of the input video signal of the terminal; The transmitter skips encoding on the detected images; Receiving the skipped frames, the receiving unit comprises the step of copying all previously received frames and playing in place of the skipped frames, The end frame detection is calculated through the derivative of the predictive PSNR obtained between the input image and the reconstructed previous image. In a video call system using a wireless terminal, A detector for detecting a start frame and an end frame of a sudden change section of an input video signal of a terminal; A transmitter for skipping frame encoding on all detected images; And a receiver configured to copy and alternately reproduce all frames previously received for the skipped frames. And the detector calculates the derivative of the predictive PSNR obtained between the input image and the reconstructed previous image.

Images