逛奔的蜗牛

我不聪明,但我会很努力

   ::  :: 新随笔 ::  ::  :: 管理 ::
1. Bilinear
2. BiCubic
3. Area_Average
4. Progressive Bilinear
当图片缩小到原图的一半以下时,bilinear的效果就不好了,再小下去,bicubic的也不够好,最好的效果是area_average算法,但是这个需要花太多的时间,可以使用改进过的bilinear算法,效果跟area average差不多,速度在bilinear和bicubic之间,但是当图片非常大,绽放到非常小时,花的时间比bicubic多得多,但是比area average少得多,效果与area average差不多,还是不错的

Progressive Bilinear Scaling

We know that a significant problem with the quality of the bilinear approach occurs when the downscale is by more than 50 percent. So what if we compen- sated for that problem by scaling iteratively toward the final size, scaling down by exactly 50 percent each time until the final iteration, when we scale by 50 percent or less? Then we would account for all of the pixels along the way that should figure into the final image. 

    // 缩小图片,缩小时可以使用改进过的bilinear, bicubic插值算法

    // 但是转换透明图片时如果使用单缓冲区会出问题,这时每次都要创建一个缓冲区才可以

    public static BufferedImage getFasterDownScaledInstance(BufferedImage img,

                                                            int targetWidth,

                                                            int targetHeight,

                                                            Object hint,

                                                            boolean progressive) {

        int type = (img.getTransparency() == Transparency.OPAQUE) ? BufferedImage.TYPE_INT_RGB

                : BufferedImage.TYPE_INT_ARGB;

        BufferedImage ret = (BufferedImage) img;

        BufferedImage scratchImage = null;

        Graphics2D g2d = null;

        int w = 0, h = 0;

        int prevW = ret.getWidth();

        int prevH = ret.getHeight();


        if (progressive) {

            // Use multistep technique: start with original size,

            // then scale down in multiple passes with drawImage()

            // until the target size is reached

            w = img.getWidth();

            h = img.getHeight();

        } else {

            // Use one-step technique: scale directly from original

            // size to target size with a single drawImage() call

            w = targetWidth;

            h = targetHeight;

        }


        do {

            if (targetWidth < img.getWidth() && progressive && w > targetWidth) {

                // 如果是缩小,宽缩小为原来的一半

                w >>>= 1;

                w = (w < targetWidth) ? targetWidth : w;

            } else {

                w = targetWidth;

            }


            if (targetHeight < img.getHeight() && progressive && h > targetHeight) {

                // 如果是缩小,高缩小为原来的一半

                h >>>= 1;

                h = (h < targetHeight) ? targetHeight : h;

            } else {

                h = targetHeight;

            }


            if (scratchImage == null) {

                // Use a single scratch buffer for all iterations

                // and then copy to the final, correctly sized image before

                // returning

                scratchImage = new BufferedImage(w, h, type);

                g2d = scratchImage.createGraphics();

            } else if (type == BufferedImage.TYPE_INT_ARGB && scratchImage != null && g2d != null) {

                // 透明图片不能使用单缓存

                g2d.dispose();

                scratchImage = new BufferedImage(w, h, type);

                g2d = scratchImage.createGraphics();

            }


            g2d.setRenderingHint(RenderingHints.KEY_INTERPOLATION, hint);

            g2d.drawImage(ret, 0, 0, w, h, 0, 0, prevW, prevH, null);

            prevW = w;

            prevH = h;

            ret = scratchImage;

        } while (w != targetWidth || h != targetHeight);


        if (g2d != null) {

            g2d.dispose();

        }


        // If we used a target size, the results into it

        if (targetWidth != ret.getWidth() || targetHeight != ret.getHeight()) {

            scratchImage = new BufferedImage(targetWidth, targetHeight, type);

            g2d = scratchImage.createGraphics();

            g2d.drawImage(ret, 0, 0, null);

            g2d.dispose();

            ret = scratchImage;

        }

        return ret;

    }

posted on 2011-01-09 18:20 逛奔的蜗牛 阅读(4116) 评论(0)  编辑 收藏 引用 所属分类: Java

只有注册用户登录后才能发表评论。
网站导航: 博客园   IT新闻   BlogJava   知识库   博问   管理