具体实施方式 Detailed ways

为让本发明的上述和其它目的、特征和优点能更明显易懂，下文特举较佳实施例，并配合附图作详细说明如下。In order to make the above and other objects, features and advantages of the present invention more comprehensible, preferred embodiments are exemplified below and described in detail with accompanying drawings.

图2所示为依照本发明一实施例的影像稳定装置及其耦接方式。在此图中，标示204即为所述的影像稳定装置，其主要包括有脸部侦测单元208及影像稳定单元210，此外，还包括有移动估计单元206。脸部侦测单元208用以对摄影单元202所撷取到的动态影像进行脸部侦测(facedetection)，以取得此影像中的一脸部区域。移动估计单元206则用以估计此影像的移动向量(motion vector)，以产生移动向量的估计结果。至于影像稳定单元210，其用以固定上述脸部区域的大小及位置，并依据移动向量的估计结果调整此影像中，脸部区域以外的其它区域，进而获取稳定的影像OUTPUT。FIG. 2 shows an image stabilization device and its coupling method according to an embodiment of the present invention. In this figure, 204 is the image stabilization device, which mainly includes a face detection unit 208 and an image stabilization unit 210 , and also includes a motion estimation unit 206 . The face detection unit 208 is used for performing face detection on the dynamic image captured by the camera unit 202 to obtain a face area in the image. The motion estimation unit 206 is used for estimating a motion vector (motion vector) of the image to generate an estimation result of the motion vector. As for the image stabilization unit 210 , it is used to fix the size and position of the above-mentioned face area, and adjust other areas in the image other than the face area according to the estimation result of the motion vector, so as to obtain a stable image OUTPUT.

在此例中，移动估计单元206估计移动向量的方式，可以是采用估计此影像的区域移动向量(local motion vector)及全域移动向量(global motionvector)的方式来实现，亦或是采用其它可行的估计方式来实现，并无特殊限定。而脸部侦测单元208所采用的脸部侦测方式，可以是利用侦测人脸的五官特征，例如侦测眼睛、鼻子、嘴巴...等特征来取得影像中的脸部区域，也可以是利用侦测颜色的特征，例如侦测肤色、发色...等特征来取得影像中的脸部区域，甚至是利用侦测纹理、形状...等其它方式来取得所需的脸部区域，也并无特殊限定。In this example, the manner of estimating the motion vector by the motion estimation unit 206 can be implemented by estimating the local motion vector (local motion vector) and the global motion vector (global motion vector) of the image, or by using other feasible methods There is no special limitation to realize the method of estimation. The face detection method adopted by the face detection unit 208 may be to use the facial features of the detected face, for example, to detect features such as eyes, nose, mouth, etc. to obtain the face area in the image, or It can use the features of color detection, such as skin color, hair color, etc. to obtain the face area in the image, or even use other methods to obtain the desired face There are no special restrictions on the external area.

至于上述的影像稳定单元210，其固定脸部区域的位置的方式，可以是将脸部区域固定在影像的画面上的原获取位置上，也可以是将脸部区域固定在影像的画面上的任何一预设位置。然而，在视讯通话的应用中，最理想的方式，就是将脸部区域固定在影像的画面中央。而影像稳定单元210固定脸部区域的大小的方式，可以是将脸部区域固定在原获取大小，也可以是将脸部区域的大小与影像的整个画面的大小固定在一预设比例，亦或者是将脸部区域的大小固定在一预设面积。举例来说，在视讯通话中，最希望看到的就是大小及位置皆稳定的人脸，因此可将影像稳定单元210设计成能将脸部区域固定在影像的画面中央，且脸部区域的大小也与整个画面维持一定比例，因此就不会有人脸抖动的情形发生。换句话说，无论视讯通话中的手机怎么晃动，此影像稳定单元210都可将偏移的脸部区域拉回画面中央，并利用缩放(zoom in/zoom out)脸部区域的方式将其大小固定住。As for the above-mentioned image stabilization unit 210, the way to fix the position of the face area can be to fix the face area on the original acquisition position on the image screen, or to fix the face area on the image screen. any default position. However, in the application of video call, the most ideal way is to fix the face area at the center of the video frame. The image stabilization unit 210 fixes the size of the face area, which can be to fix the face area at the original acquisition size, or to fix the size of the face area and the size of the entire frame of the image at a preset ratio, or It is to fix the size of the face area at a preset area. For example, in a video call, what you want to see most is a face with stable size and position, so the image stabilization unit 210 can be designed to fix the face area in the center of the image, and the face area The size also maintains a certain ratio with the whole picture, so there will be no shaking of human faces. In other words, no matter how the mobile phone shakes during the video call, the image stabilization unit 210 can pull the offset face area back to the center of the screen, and zoom in/zoom out the face area to reduce its size. fix.

此外，当脸部侦测单元208对上述影像进行脸部侦测，因而取得多个脸部区域时，那么影像稳定单元210便会选择这些脸部区域的其中的一个来作为固定对象，而其余的脸部区域则视为背景。而影像稳定单元210选择固定对象的方式，可以是选择最大的脸部区域、于画面最中央的脸部区域或符合一预设脸孔的脸部区域来作为固定对象，也可以是依据其它的预设条件来选择其中一脸部区域来作为固定对象。In addition, when the face detection unit 208 performs face detection on the above-mentioned image to obtain multiple face regions, the image stabilization unit 210 will select one of these face regions as a fixed object, and the rest The face area of , is regarded as the background. The way for the image stabilization unit 210 to select the fixed object may be to select the largest facial area, the facial area in the center of the screen, or a facial area matching a preset face as the fixed object, or it may be based on other factors. Preset conditions to select one of the facial areas as a fixed object.

利用上述操作，可以归纳出一些基本的操作步骤，如图3所示。图3所示为依照本发明一实施例的影像稳定方法的流程。在此方法中，首先是对一影像进行脸部侦测，以取得此影像中的一脸部区域(如步骤S302所示)。接着，固定上述脸部区域的大小及位置(如步骤S304所示)。如此一来，便可稳定此影像中的脸部区域，解决影像的人脸抖动问题。若欲取得更佳的影像质量，则可再估计此影像的移动向量，并根据移动向量的估计结果调整上述影像中，脸部区域以外的其它区域(如步骤S306所示)。Using the above operations, some basic operation steps can be summarized, as shown in Figure 3. FIG. 3 is a flowchart of an image stabilization method according to an embodiment of the invention. In this method, firstly, face detection is performed on an image to obtain a face area in the image (as shown in step S302). Next, the size and position of the aforementioned face area are fixed (as shown in step S304). In this way, the face area in the image can be stabilized to solve the face shaking problem in the image. If better image quality is desired, then the motion vector of the image can be re-estimated, and other areas in the above image other than the face area can be adjusted according to the estimation result of the motion vector (as shown in step S306 ).

若欲传送拍摄到的影像，则其基本操作步骤如图4所示。图4所示为依照本发明一实施例的影像传送方法的流程。在此方法中，首先也是对一影像进行脸部侦测，以取得此影像中的一脸部区域(如步骤S402所示)。接着，固定上述脸部区域的大小及位置(如步骤S404所示)。然后，压缩上述已固定脸部区域的影像，以传送至一远程(如步骤S406所示)。如此一来，便可在传送影像之前，先行稳定此影像中的脸部区域，解决影像的人脸抖动问题。当然，若欲取得更佳的影像质量，则可在压缩上述已固定脸部区域的影像前，估计上述影像的移动向量，并根据移动向量的估计结果调整影像中，脸部区域以外的其它区域。If you want to send the captured images, the basic operation steps are shown in Figure 4. FIG. 4 is a flowchart of an image transmission method according to an embodiment of the present invention. In this method, face detection is firstly performed on an image to obtain a face area in the image (as shown in step S402). Next, the size and position of the aforementioned face area are fixed (as shown in step S404). Then, compress the image of the above-mentioned fixed face area to send to a remote (as shown in step S406). In this way, before the image is transmitted, the face area in the image can be stabilized first, so as to solve the face shaking problem of the image. Of course, if you want to obtain better image quality, you can estimate the motion vector of the above image before compressing the image with the fixed face area, and adjust the other areas in the image other than the face area according to the estimation result of the motion vector .

若欲接收对方所拍摄到的影像，然对方却没有稳定影像中的脸部区域的功能，则其基本操作步骤如图5所示。图5所示为依照本发明一实施例的影像接收方法的流程。在此方法中，首先是接收并解压缩一影像压缩数据，以取得一影像(如步骤S502所示)。接着，对此影像进行脸部侦测，以取得影像中的一脸部区域(如步骤S504所示)。然后，固定脸部区域的大小及位置，以获得一显示影像(如步骤S506所示)。如此一来，便可在取得对方所传送的影像之后，先行稳定此影像中的脸部区域，解决影像的人脸抖动问题，然后再显示出来。当然，若欲取得较佳的影像，则可在固定脸部区域的大小及位置时，估计上述影像的移动向量，并根据移动向量的估计结果调整影像中，脸部区域以外的其它区域，以获得所需的显示影像。If you want to receive the image captured by the other party, but the other party does not have the function of stabilizing the face area in the image, the basic operation steps are shown in FIG. 5 . FIG. 5 is a flowchart of an image receiving method according to an embodiment of the invention. In this method, firstly, an image compression data is received and decompressed to obtain an image (as shown in step S502). Next, face detection is performed on the image to obtain a face area in the image (as shown in step S504). Then, the size and position of the face area are fixed to obtain a display image (as shown in step S506). In this way, after obtaining the image sent by the other party, the face area in the image can be stabilized first to solve the face shaking problem of the image, and then displayed. Of course, if you want to obtain a better image, you can estimate the motion vector of the above-mentioned image when the size and position of the face area are fixed, and adjust the other areas in the image other than the face area according to the estimation result of the motion vector, so as to Get the desired display image.

综上所述，本发明因对影像进行脸部侦测，以取得此影像中的一脸部区域，然后再固定此脸部区域的大小及位置，故可稳定影像中的脸部区域，解决影像的人脸抖动问题。此外，若是再搭配估计上述影像的移动向量，并根据移动向量的估计结果调整影像中，脸部区域以外的其它区域，便可得到更佳的影像。当然，本发明除了可应用在手机、其它具有摄影功能的手持式装置及车载摄影机之类的电子产品之外，亦可应用在其它需要实时传送影像的相关领域。In summary, the present invention detects the face of the image to obtain a face area in the image, and then fixes the size and position of the face area, so the face area in the image can be stabilized, solving the problem of Face shake problem in video. In addition, a better image can be obtained by estimating the motion vector of the above-mentioned image and adjusting other areas in the image other than the face area according to the estimation result of the motion vector. Certainly, the present invention can be applied to electronic products such as mobile phones, other handheld devices with camera functions, and vehicle-mounted cameras, and can also be applied to other related fields that require real-time transmission of images.

虽然本发明已以较佳实施例描述如上，然其并非用以限定本发明，本领域技术人员在不脱离本发明的精神和范围内，当可作些许得更动与润饰，因此本发明的保护范围应当以权利要求书所界定的内容为准。Although the present invention has been described above with preferred embodiments, it is not intended to limit the present invention. Those skilled in the art may make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, the present invention The scope of protection should be determined by the content defined in the claims.