Micro scan code-why so fast? Official science-micro letters, scan code, QR code-IT information
IT information on QR code scanning has become the most popular social networking links, network access, mobile media, brushed myself a micro letter help app to quickly enhance customer quantity and frequency of use. This looks easy function there is no secret behind it? Micro-credit technology brings the official science team today, details of the probe into a QR code scanning technology.
QR code scanning feature is not the beginning, of course appear in tiny letters, did not join until app 5.0. Scanning code has been used to add friends, access to business brand, or even a single product information as well as micro-credit payment, eliminating the users look for information online when you search, also gives users a shortcut to connect online and offline resources, users do not need to download a QR code reading applications.
Micro scan code-why so fast?
QR first map:
QR shape is a square with three corners on the "return" Word identification, is used to identify two-dimensional code key positioning graphics, or position detection graph (pictured below).
Scan code process is continuous AF, image collection, identification process acquisition less the number of frames on behalf of scan codes faster, can be understood as a graphics processor to refresh a few times per second.
Micro-team analysis of data shows that engine added to the normal picture decoding speediOS can be shortened to 5ms, Android is only approximately 12ms.
Addition app scan code there is Prediction algorithm, it acts as a mine-clearance point, before proceeding to the core processing, can step back to a large number of images in the "code" investigative judgment.
Whole sweep code pre sentenced module is located in core recognition engine zhiqian, no longer need on entered of video in the of each frame image for detection recognition, can achieved fast filter large uncensored image, reduced follow-up not necessary of positioning and recognition on sweep code client caused of blocked, makes response more timely, increased sweep code process in the of smooth degrees, such, everyone using up to in "no perception" of State Xia experience micro-letter sweep code of fast.
Earlier tests showed uncensored images on mainstream mobile processing speed has greatly improved, some models average running time is reduced to less than 10ms. For code image, two-dimensional code recognition is also close to 35%, and in the case of time keep the rate unchanged.
In addition to the above app QR quick scanning, micro-sweep coding accuracy down the effort.
Micro letter team developed a two-dimensional code and barcode recognition "the killing machine"-recognition engine QBar. QBar scan code for two-dimensional codes fault-tolerance decoding algorithm, multiple language character detection algorithm using dozens of optimization, recognition rate and speed has been improved.
微信扫码为什么那么快？官方科普来了 -
微信,扫码,二维码 - IT资讯
IT资讯讯 二维码扫描已经成为当下最流行的社交联系、网络访问、移动支付等媒介，而微信中的扫一扫也帮助微信迅速提升用户量和使用频次。这项看起来简单方便功能的背后有什么秘密呢？微信技术团队今天带来了官方科普，一探微信二维码扫描技术细节。
当然二维码扫描功能并非一开始就出现在微信上，直到微信5.0才加入。扫码已用于添加好友、获取商家品牌、甚至单个商品信息以及微信支付，免除了用户在寻找信息时的在线搜索过程，也给用户指引了一条连接线上线下资源的捷径，用户不需要另外下载一个二维码读码应用。
微信扫码为什么那么快？
先来看看二维码结构图：
二维码外形为正方形,三个角方向上有“回”字标识，是用来识别二维码的关键定位图形，或者叫位置探测图形（如下图）。
扫码过程其实就是不断对焦，采集图像，识别的过程；采集的帧数越少，代表扫码速度越快，可以理解为图形处理器每秒钟能够刷新几次。
微信团队分析数据显示，引擎在识别正常图片时的解码速度，iOS可缩短至5ms， Android也仅仅约12ms。
另外微信扫码还有预判算法，它充当扫雷尖兵，在进行核心识别处理之前，能早一步对大量图像中是否有“码”进行侦查判断。
整个扫码预判模块位于核心识别引擎之前，不再需要对输入的视频中的每一帧图像进行检测识别，能实现快速过滤大量无码图像，减少后续不必要的定位和识别对扫码客户端造成的阻塞，使响应更加及时，增加扫码过程中的流畅度，这样，大家使用起来才能在“无感知”的状态下体验微信扫码的快速。
更早前测试显示，对于无码图像，主流手机上的处理速度均有较大幅度地提高，部分机型平均运行时间降低到10ms以内。对于有码图像，二维码的平均识别也快了近35%，而在时间缩短的情况下保持了识别率不变。
除了以上微信二维码快速扫描外，微信在扫码准确性上还下了一番功夫。
微信团队专门开发了二维码和条形码识别“大杀器”——识别引擎QBar。QBar扫码对二维码容错性解码算法、多语言字符检测算法等均使用进行了数十项优化，在识别率和识别速度上得到了提升。
