AF方案调研
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290fefbaae704b34f7bc4afea9c6f48a417021d93cef6b86d450a72d48f387cf157b14b71aade342f53e37c7cbe804a5857f154bc80ff229874acbd56e4f25b13a7e19054e5be5287cdde496eff16fc0202c24fef1af81c16a87cb6aecd627ed3ec6036d1c0e1af6f9b5b4e003f2226a10fdbf928488f6da05b37a666d456474c87558792cac3239ea888d8586dfad89a71008dad6556052b757f9f63de7671b61aca6dbf58009d3fe48359cc57fb8a5491324d71ff342ae1b31095e416b33c25a41ab823a3131c75ae8d70eb8e8ad2b5353dc2cabaea1adc64ae ...
预录方案
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290f759b998e57fbbecb5fc9d690e98d1bef20e40941bf35a97652281ac3cfb72a22c7a6c27b6bb82b6d991197602215612de101f97ca5db7e14e54e64ce7ed93074d2db7614f9d983b78ee72e19a3df5993bf812d2213332d554862458052a906a36dc90a4b3597d3b9806b65aafc84e705e350d8c19cb3a72644403d96ea7e15539528a4a53625cd6d96b36e01babc6a190d1e5fe3904a3052b9cd2c69bebf641bf4ab1007b8c6cba6c91779f9c1ea852b0bb715ac26db3ad9e5438cd37288d2d4e6f080b7f9f49ebb405d098089ce696cf1c78df66f7665cba ...
相机自动曝光
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290f61b946f1556bc4bf333be9d816f2defbfce7eb189e9960dd56298a09c6b2d44938892f67539e9a242aa4d9498f562b0a1d48c39175c2688cf1a65a191822eeed559e855fb701f200dc5d7abb3b93fcb72f4ccbab113107c76789ecc81b412950cda2daa4d03f5f1bcdc6c2a1c1f600de3fbfc425541817ecd046b5c097f8d70bd3824d8cea7ed2a204bf1b603e458edd2385ea9f21debbcb80173f3e1003e6c9f006ce41c4a44884341c5a8d7c88577bcc76b338411dad9fe177e36d4dc04d0f7eb8522fcc86baaf842a7e1af884d97bbe1609f755bfe1270 ...
相机模组的图像校正及测试方法
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290fac8306cb27687a7bcf5c3c1c53bd33ccb380e6f8aaf1cd4af51f45ce18a2086c9a935c8b0dbc82cffcb288dc1173de8588fbc1ae8c87e1b821c1fe126a845d4d2564392606f4fc9c1c3a10cf1b763d4bf6d36755ba20f840f54c462d811505027416d9e0395f3d39154a911bd75616482f991a064d3a14c2ad224e4b7b5563a44db982cea7bf382d91bfbeb532642f7c09344db25c8473cf0d5f1555ada2ce71b16e13f0987ccfa7b4ea5562d572d78339969df338a4c65421cf2bf654cc5293b37deb0a198ead98387068ac6525fbedfed1400254b4e9753 ...
文件尾在JPGRAW（DNG）中的存储结构
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290fdc908339a4ceb1ba1556f0cb58485611bd57a9b58ddbfd0109c3dc89337cd774a6bd91f266f4df62e90632151fb1502afc443e50feffe3b398b6a82d5a226d8ed180e0b24d72d9de5b345c5dd416139e7f515b7b55534584a67d15f0d30d1011c6f38fdd324f2e5736e82fce7e464fe22a96d21b0b3969158d364209b3de1e4305d1a9651eb88c758ded664975444788f6614a648f1e907f1bb53ffdf38059680eeccafabe9d8d57250fb67b24473c72776bf254e5fd71000f594cd718c67f883dafc502ac1464295a9f359d8deb4b2b16ae90cfab3c663b8 ...
CV Five 流程
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290f6a512d408f0f2fa4d62ed7120d0d65d53b92f45045d963089e2b7ce4ca3911453e88f0a07f21286dadde7eef685d906f4d072d43487164ef95ad0bc5a2c51c26282184c7de445c636f53cb6e91cb4fc80e544983bdffcab659c7cd634240df71a77b614a8262f13248082bf94b64013f8f66c28c432cc9322ba6a4fdc2aab2be940615c7fc66b5658bc5a8ae9ceef5382ef3614b039372bb8c11936fd152c05ad591713fb5f9b68b181baade4d819ca16d1be1b79ba3478ea683a6217891503d8e029c96500932fe0b058513f153a6fd0d1e30df1a68d750c ...
色彩一致性验收标准
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290f6c4dfce00379a6f48d91d0c29bf36a11a4ebdeb6627e64f161b9a63c8f5bf5413bbaeca4a1a8538cd8de717a1ed0466a99fe8ac866d6a780ef43ef75275505a2dbce169a911c9d3d11d67f2b348562dd59fa91f87d2e76749fd9d85bd35e97c32e21f099eea5ce43ffcc3b3e4855691449515597dfb92d455dce028e7e373672bc8970aaf06190e6472746dfbe19847e76c97e3ef4e6ed3b06b14b7907f0020885e8e5a1528de56cdbbdac27f3be44b6866086960ca47937a2731c38a8c98743511c306eb92052d6f136b2125c7540cce4f7d4b3820cb8d1c ...
色彩调试LUT
ff1e8db80addafe01df7f38ed8efc76b8d3928dcd6686de7ef1720eeb0419c8efe7a9ba9ac9506fed8a9f26220b63245449f9dffa182d09502e2e090440a77020284733c89f140ae2edc0af7c9935284fc75dedaa42b392566bd5b6b24060f29bec5130905d8ef13fb06b8c1fabd1928f944bece095025392f15fa118544065e121bc4a6eac8a5c1895c7b0f1689bdd2e2f068c3a65e14d79fbf6a0a266fefd55154d8f85d287fc4baf2bd654bb0effbc868e711f72deee31596081127b9503dda0bb02e90487ffb8406ea6493afc1e55487bba992b6b39b92941f3b99239fc0f2add8df1a37e1c57537f8e08a5be2c4750e7092660ea6287 ...
录中拍流程
ff1e8db80addafe01df7f38ed8efc76b88584c8b687473db66c159aa9c47290fc4d9bad0fefae7919b5d4ee957dd8ef76edf5145afbff372c8643aa5e0bf44af92cec58a2c6e995fc174438c97c3f94ab5d3b3c4be0666b926d44a34d52600af00bbea1447b25ec22e372d4bcd1aac18553af30226940170288a997bd8ee5a632010ab45cf4200e443793630dd6344a5e70c277f844b50ae8a5681e462675c7b186f33661819f558487ae477266eb6b1cd64de12e0ce9b99736a19e8a2c67caf28b34c72ee82dddb5dbc37bc8e67db9d93b030501d0a94238f87b4b32951595b6439ba2e2fdc941864a72e1b189f0b3929b0348caf3c92491 ...
ISP 之 Pipeline
什么是 ISP
主流的 CMOS 和 CCD sensor 几乎都是输出 Bayer mosaic 格式的 RAW
数据，这种数据格式是无法直接观看的，必须转换成常见的 RGB 或 YUV
格式才能被主流的图像处理软件支持。对于 camera 产品而言，一般还需要将 RGB
或 YUV 图像进一步转换成 JPEG
格式以方便进行存储。上述图像处理过程统称图像信号处理（Image Signal
Processing，ISP），广义的 ISP 包含了 JPEG 和 H.264/265
图像压缩处理，而狭义的 ISP 仅包括从 RAW 格式变换到 RGB 或 YUV
的处理过程。
一个典型的 ISP 流水线由一系列处理模块组成，这些模块首尾相连，在几百
MHz
的时钟驱动下同时高速运转，图像数据不断从一个模块转移至下一个模块，直到完成所有的算法处理，最终以
YUV 或 RGB 的形式从流水线的末级流出
ISP。下图所示的是一个支持常见基本功能的 ISP 流水线。
从图中可以看到，图像数据在 ISP
内部经历了两次颜色空间变换，第一次变换发生在 Demosaic 模块，它把像素从即
RAW ...