什么是影像压缩技术的极限 - 丁建均 - national taiwan university.ppt

* 聲音檔：.tw/air.mp3 time-frequency analysis * 目前的成果：20秒長度的哼歌，辨識成功率為 100% 但仍有不少精益求精的空間 * 9. Wavelet Transform 只將頻譜分為「低頻」和「高頻」兩個部分，大幅簡化了 Fourier transform (對 2-D 的影像，則分為四個部分) x[n] g[n] ? 2 x1,L[n] x1,H[n] ? 2 h[n] 「低頻」部分 「高頻」部分 Example: g[n] = [1, 1], h[n] = [1, -1] or * x[m, n] g[n] h[n] ? 2 ? 2 along n along n v1,L[m, n] v1,H[m, n] g[m] h[m] along m ? 2 x1,L[m, n] ? 2 along m x1,H1[m, n] g[m] h[m] along m along m ? 2 ? 2 x1,H2[m, n] x1,H3[m, n] 2-D 的情形 m 低頻, n 低頻 m 高頻, n 低頻 m 低頻, n 高頻 m 高頻, n 高頻 L-points L-points M ×N n m * The result of the wavelet transform for a 2-D image lowpass for x lowpass for y lowpass for x highpass for y highpass for x lowpass for y highpass for x highpass for y * -- JPEG 2000 (image compression) -- filter design -- edge and corner detection -- pattern recognition -- biomedical engineering Applications for Wavelets * 10. Integer Transform Discrete Fourier transform (DFT): Discrete cosine transform (DCT): RGB to YCbCr Transform 如果它們的 entries 都是整數 (或是 C/2b) 該有多好 DFT, DCT 很常用，但是，他們都是無理數….. * Integer Transform: The discrete linear operation whose entries are summations of 2k. , ak = 0 or 1 or , C is an integer. * Problem: Most of the discrete transforms are non-integer ones. DFT, DCT, Karhunen-Loeve transform, RGB to YCbCr color transform To implement them exactly, we should use floating-point processor To implement them by fixed-point processor, we should approximate it by an integer transform. However, after approximation, the reversibility property is always lost. * Walsh transform (applied by CDMA) * Integer RGB to YCbCr Transform This is used in JPEG 2000. * [Integer Transform Conversion]:  Converting all the non-integer transform into an integer transform that achieve the following 6 Goals: A, A-1: original non-integer transform pair, B, B?: integer transform pair