光場（Light Field） 為一個描述三維空間中所有光線行進的位置與方向的函數。對於那些由不同位置及角度觀察時顏色會隨著變化的物體，將光場參數化在物體表面上則自然的提供了一個不錯的方法來紀錄下物體表面由各角度觀察的變化。而得力於日漸成熟的顯示卡硬體加速，加上利用PCA(Principle Component Analysis) 將龐大的四維表面光場資料分解成數項Surface Map和View Map相乘的和來逼近原本的資料，Light Field Mapping能在即時的更新速率下畫出一物體的表面光場。
這篇論文則在Light Field Mapping的架構下,嘗試將Light Field Mapping用PCA分解出的View Map作處理。由於二維的View Map保留了原本四維光場中對於光線射出方向描述的參數，其資訊似乎也跟光場依據觀察角度改變的特性相關。而在真實世界中，物體反射到相機位置的光線是取決於入射光及物體的表面反射特性。所以我們嘗試將View Map根據表面法向量鏡射回光源方向以重新參數化。得到的Light Map類似光源與物體表面反射特性的convolution。在較單純的光源環境以及固定的物體表面反射特性下， 物體各處所求得的Light Map由於為光源與物體表面反射特性的convolution，所以比起原本以View Map的儲存方式擁有較高的空間相關性。藉此我們能夠以較高的壓縮效率來對其壓縮。

Light field is 4D function descript the flow of light through 3D space. A light field parameterized on the surface offers a natural an intuitive description of the view-dependent appearance of scenes with complex reflectance properties. Light field mapping extract the 4D surface light field data into several terms of view map and surface map to reduce the size. The light field data can be rendered at interactive rates with help of modern graphics hardware.
Under the framework of light field mapping, we try to process the view maps extracted from the surface light field data. Since the 2D view maps preserve the light direction part of 4D surface light field parameters, it contains the information about view-dependent property of light field. Actually, the amount of reflected light received by camera depends on the amount of incident light and surface reflectance properties. We try to reflect the view maps around the surface normal to light maps, which can be consider as convolution of light source and surface reflectance properties. Under simple light condition and uniform surface reflectance properties, each light map got from each vertex should be similar. Because of the data have more special coherence, we can compress the data in more efficient way.

[Bishop95] C. M. Bishop. Neural Networks for Pattern Recognition. Clarendon Press, 1995.

[Chen02] Wei-Chao Chen ， J. Bouguet , Michael H. Chu and Radek Grzeszczuk. Light Field Mapping: efficient Representation and Hardware Rendering of surface Light Fields. In SIGGRAPH02, pages 447-456, July 2002.

[Gersho92] A. Gersho and R. M. Gray. Vector Quantization and Signal Compression. Kluwer Academic Publishers, 1992.

[Glorter96] Steven J. Gortler, Radek Grzeszczuk, Richard Szeliski, and Michael F. Cohen. The lumigraph. SIGGRAPH 96, pages 43–54.

[Lee99] D. D. Lee and H. S. Seung. Learning the Parts of Objects by Non-Negative Matrix Factorization. Nature, 401:788–791, 1999.

[Levoy96] Levoy and P. Hanrahan. Light Field Rendering. Proceedings of SIGGRAPH 96, pages 31-42, August 1996.

[Miller98] G. S. P. Miller, S. Rubin, and D. Ponceleon. Lazy Decompression of Surface Light Fields for Precomputed Global Illumination. Eurographics Rendering Workshop 1998, pages 281–292, June 1998.

[Wood00] D. N. Wood, D. I. Azuma, K. Aldinger, B. Curless, T. Duchamp, D. H. Salesin, and W. Stuetzle. Surface Light Fields for 3D Photography. Proceedings of SIGGRAPH 2000, pages 287–296, July 2000.