You'd need to look in the html source of the epub and see what tags and attributes are being used that will uniquely identify the chapter headers (usually <h1>, h2, h3, h4 or <div class="chapter"> and the like). Armed with that info, you can use the 'wizard' button to build the regex for you. You can use the Tweak epub feature (right click on the ebook in Calibre) to explode and explore the epub source. You won't need to rebuild the epub if you're just looking.
