Values returned by read_imagef in OpenCL are in the range of 0 to 1, while the pixel intensity range 0 to 255. .. I want to process the intensity values for image segmentation which is based only on intensity values (the operation is done to find differences between two intensity values, sorting, finding the minimum, shuffling e.t.c). How to get these values in the integer form of 0 to 255 (for convenience) ...? Can I multiply every value with 255 or do I have to do something else ...? read_imagef float4 , And channel data types can be used for images created with UNORM_INT8, UNORM_INT16, SNORM_INT8, SNORM_INT16, HALF_FLOAT and Flat. The conversions are part of the rule, and are given in section 8.3.1. To answer your question, yes, if your image data type is UNORM_INT8, then you can multiply only by 255. GPUs are usually faster with UNORM_INT8 with UNSIGNED_INT8 images.
No comments:
Post a Comment